Internal Quality Control Program Panel 1

Principles and Criteria 4

Saliva Testing 6

Hair Testing 12

Sweat Testing 20

Urine Testing 29

Onsite Testing 41

Questions and Answers 50

Reporting Test Results Panel 86

Principles and Criteria 86

Saliva Testing 90

Sweat Testing 94

Onsite Testing 119

Interpreting Test Results Panel 161

Principles and Criteria 163

Hair Testing 165

Urine Testing 174

Saliva Testing 180

Onsite Testing 189

External Quality Assurance Program Panel 243

Principles and Criteria 245

Urine Testing 248

Hair Testing 257

Onsite Testing 264

Saliva Testing 275

Questions and Comments 286

DR. BUSH: Good morning. It is so good to see so many bright, fresh faces at 8:15. I told you yesterday that we have quite a challenging day. Let's see how we all look by about 5:30 this afternoon.

[Discussion off record.]

DR. BUSH: The purpose of this meeting is to

provide critical and new information to help the Drug Testing Advisory Board make recommendations to us concerning the use of alternative specimens and technology for drug testing in the workplace. The drug testing advisory board will be working with the information derived from this meeting. We need all of the presenters to provide the best possible responses to the Drug Testing Advisory Board questions.

While it may not be possible for each presenter to provide a full and detailed answer at the time the question is asked, we request that the Drug Testing Advisory Board members keep track of their information needs and make them clearly known to the presenters.

As a follow-up to that, yesterday afternoon we

needed one slide one slide from Dr. Armbruster. He gave it to us right there on the spot. Dr. Niedbala provided a couple of slides that somehow, through somebody's oversight or whatever, just did not get included in this group. We want the completeness for the final document. Dr. Kippenburger, and Dr. Baumgartner have agreed to have a transcription of Dr. Baumgartner's presentation prepared with the slides that he presented in order with that presentation so it may be clearer to us some of that information and how everything meshes together. So those needs we believe at this time have been voiced and met. But, please, all of you, work together so that the best possible information can be provided us, provided the Drug Testing Advisory Board to make recommendations to us. We thank you for that.

When I came in here a little earlier, there were three people signed up to make public comments tomorrow. Take the opportunity should you want to do this. This is your arena to do it. So, please sign up should you have comments.

We are going to take a look at internal quality control programs this morning. Dr. Mike Baylor chairs this panel. Dr. Mike Baylor received his Ph.D. degree in pharmacology and toxicology from West Virginia. We read this before, but I do know we have some new people in the audience this morning, so we will go through it again. I am taking some flack on my CV reading resume reading. He immediately began his forensic career in a clinical toxicology post-mortem tox lab in Germany. Since then, he has been involved in over 20 years in various aspects of forensic toxicology and drug testing.

Dr. Baylor was the Director of the U.S. Army Drug Testing Laboratory at Fort Meade, Maryland from 1985 to 1987. For those of you who do not know, he hired me when I came into that laboratory.

In 1987, he left the military to become the

Technical Director of one of the largest commercial drug testing laboratories located in Research Triangle Park, North Carolina, testing over 4,500 specimens a day. He has been an active NLCP inspector since 1988.

Dr. Baylor was appointed as a Senior Staff Fellow recipient with the National Institute on Drug Abuse from 1990 to 1993 and has been a special expert with SAMHSA from 1993 to 1994. That is when he worked for me. He played a major role in developing technical policies and procedures for workplace urine drug testing for the NLCP and numerous federal agencies.

Dr. Baylor is currently the senior research

forensic toxicologist at the Research Triangle Institute in Research Triangle Park, North Carolina, where he is one of the key staff who are responsible for the operation of the National Lab Certification Program. His primary focus is in the inspection process and remediation. Dr. Baylor has accumulated a wealth of experience and knowledge in workplace drug testing over the past two decades and, hence, he is chairing this session. Dr. Baylor.

Agenda Item: Internal Quality Control Program

Panel

DR. BAYLOR: There was a time when I used to say, Dr. Bush, just do it.

[Laughter.]

DR. BAYLOR: And then there was a time when I said, yes ma'am, I will do it.

[Laughter.]

DR. BAYLOR: Times sure change. This morning's session is going to address the forensic and scientific

requirements for internal quality control and assurance. As indicated in Dr. Cone's presentation yesterday, the presenters this morning will address how quality control and assurance are handled in the alternate matrices.

Some of the focuses that should be -- that will be focused on in this morning's presentation are the certified reference standards that are used in the different analyses, appropriate control materials, the use of open controls, as well as blind controls, performance testing, quality assurance, as far as the documentation and review process, in addition to the certification results that would be initiated from the analysis of the specimens.

In keeping on schedule, this morning's first

speaker will be Dr. Michael Peat. He joined Lab One in June of 1994 as a Senior Vice President for Toxicology. Since May of 1996, he has served as Executive Vice President of Toxicology. He is responsible for the operation of the Toxicology Laboratory, as well as the Substance Abuse Testing Sales Force and Client Services Department.

Prior to his employment at Lab One, Incorporated, Dr. Peat was Vice President of Toxicology for Roche Biomedical Laboratories, as well as Compuchem Laboratories previous to that.

His previous experience includes appointments at the Metropolitan Police Forensic Sciences Laboratory, near Scotland Yard, London, and the Center for Human Toxicology, the University of Utah, Salt Lake City. He has an undergraduate degree in chemistry, a doctorate in pharmacology, and is certified by the American Board of Forensic Toxicology, where he presently serves as Director.

He is also currently the President of the American Academy of Forensic Sciences, which was left out of his resume yesterday.

Dr. Peat, until October 1994, was a member of NIDA's Drug Testing Advisory Board, with the Department of Health and Human Services, and is a consultant to the Toxicology Resource Committee of the College of American Pathologists. He is the author of more than 40 peerreviewed articles and eight book chapters. Dr. Peat.

DR. PEAT: Thank you, Mike, and good morning.

Before we start, just an historical perspective on quality control. Some of you in the audience, and this will

probably date you, may remember the meeting in Toronto where were discussed quality control in the forensic toxicology arena. It is probably at least 10-15 years ago that we had that meeting. At that meeting, a dear, respected toxicologist said: "What do we need accurate quantitation for? All we need to know is whether it is a little, a lot, or a hell of a lot."

[Laughter.]

DR. PEAT: We have come a long way since that time because we now recognize that there is a need for accurate quantitation.

I am going to talk a little bit about saliva.

But, before I do, I would like to just mention some of the - what is the basis obviously of an internal quality control program. Certainly, the idea behind that or the mechanism behind that is to ensure accurate and reliable data.

data. The means by which that is done is obviously the monitoring of the testing data, interpretation of that data, corrective action taken, if that data shows that the assay is out of control or trending to out of control, and then documentation of all of that.

Really, the principles are the same, irrespective

of the specimen that you are testing. Certainly, when we talk about saliva, we can very similarly equate that to urine. The urine quality control guidelines have evolved through the inspection process, NLCP inspection process. So that now, on an initial screen, the generally-accepted practice is to run negative controls, 75 and 125 percent controls and to monitor those data for positive and negativity.

On the confirmation, it is to run a control above the cut-off, generally at 125 percent of cut-off and one that is below the cut-off. Certainly, for retest purposes, it is recommended that a 40 percent control be analyzed.

The issue really is, as you look at some of these alternative specimens, can the range of certified labs produce data that are accurately reliable at some of the concentrations that have been discussed. If you take cocaine, metabolite benzoylecgonine, and Tim Rohrig's discussion on saliva yesterday, when you talk about a 40 percent control, you are talking somewhere in the five nanogram per milliliter for benzoylecgonine. That is on a very small volume of fluid. We are not talking of having five mLs or 10 milliliters of fluid available. We are talking of having a half a milliliter or maybe one milliliter available.

So, the question really is can the range of certified lab produce those results accurately and reliably?

The answer probably at this point is that, across-theboard, it would be very difficult for a number of them to do that. Certainly, when you look at THC and other drugs, those concentrations may be in fact lower.

So we certainly have similarities to urine testing as far as saliva is concerned. The analytical protocols are very similar. As I said, lower volumes are available. The mix of immunoassays and MS techniques may differ. There may be a need for more sophisticated MS techniques than we currently use.

Obviously, the applying of QC techniques in a forensic environment have caused some issues in the past, but, generally, those have been resolved through the inspection process at this point.

Certainly, when you look at a lot of statistical data, it is important to bear in mind, as you go down and down in terms of approaching the limit of detection, your confidence level, your degree of confidence in the quantitation is going to broaden and you are going to have more unreliable quantitation results at that particular time.

Do we shift slides there? What are we doing here?

I was in Brazil once and, unbeknownst to me the person in front said -- the projectionist spoke only German and the person in front said when I wave my left hand, move the slides. Well, for those of you who watch me speak on numerous occasions, I wave my left hand a lot.

[Laughter.]

DR. PEAT: We had a movie.

[Laughter.]

DR. PEAT: So, certainly, applying QC techniques

in the forensic environment is certainly not a problem these days. It is one that we have learned to live with and certainly one that should be acceptable.

When you look at saliva, it is a different

specimen matrix. Does that impact the preparation of the controls? Somewhat. But you also heard yesterday that those pads that are used are put in a preservative fluid. The preparation of control material is relatively straightforward. You just prepare that in the preservative fluid

rather than try and prepare that in the spit. You would prepare it in the preservative fluid.

Obviously, as I said, there are a couple of occasions that the specimen volume was less and the target analytes may be different. THC, for example, would be the target analyte for marijuana use. For those of you who have tried to determine THC in bodily fluids, it is a lot more difficult than trying to determine THC acid in bodily fluids. So, again, there will be some changes that come there, as with the target concentrations and the cut-off concentrations will be lower in this fluid.

There are two issues, again, that apply

generically to some of these alternative fluids, and they were sort of approached yesterday. The specimen volume is low and there may be an inadequate volume for a retest. There may be also inadequate volumes in one collection for multiple drug determinations. If you have a THC and a cocaine positive, you may not have enough fluid from one saliva collection to go ahead and do both drugs in that saliva fluid.

I asked Sam yesterday, and he was unaware if

anybody had tried this, but obviously you can go ahead and do repeatable collections of the pad. You could put that pad in the mouth, collect and then re-collect over a period of time. It is questionable whether there would be significant differences in that, but Sam did not know if anybody had tried that. Certainly, that would be something that would be needed to be tried if you talk about using the saliva fluid in this type of environment, also for the practical purpose of a split collection or a retest at the request of the donor.

The common procedure in urine drug testing is to validate the control material using GC/MS. You obviously could do the same thing with saliva and the preservative fluid. Stability data would be needed, but I have no reason to question that the analytes concerned would in fact be stable. But the degree of difficulty increases dramatically when you are talking about low nanogram per mL determinations and maybe picogram gram per mL determinations of drugs and or metabolites.

Running GC/MS/MS and running LC/MS/MS, running sophisticated techniques such as that are not as straightforward as running mass-selective detectors. For those in the audience who have used them, you will certainly be well aware of that. For those in the audience who have not, you are going to suffer some pain I think when you try to do that. So the degree of difficulty certainly increases.

Yes, it is obviously possible to prepare blind

specimens, much the same as you prepare open controls in the preservative fluid and to ship those from the collection sites in the collection devices. That is not a problem with saliva specimens, and certainly it would be very similar in design to the urine blind program that is in place today.

that used in urine programs. The major differences would be with the target analytes and the cut-off concentrations that would be necessary to be targeted with an internal qualitycontrol program. Thank you.

[Applause.]

DR. BAYLOR: Thank you, Dr. Peat. Our next

speaker will be Ann Marie Gordon. Ann Marie Gordon received her master's degree in microbiology and immunology from the University of California, at Berkeley. She has more than 10 years of research experience in microbiology and molecular genetics. She has worked as a certifying scientist and expert witness for the U.S. Army Forensic Toxicology Drug

Testing Laboratories in both Wiesbaden, Germany, and Fort Meade, Maryland for six years.

For the past three years, Ann Marie has been the Director of Quality Assurance at Psychomedics Corporation in Culver City, California.

She will be addressing internal quality assurance and quality control in the drug testing laboratory.

Agenda Item: Hair Testing

MS. GORDON: I am going to talk this morning about the Internal QA/QC Program at the Psychomedics Drug Testing Laboratory. The procedures that we have in place are not unsimilar to those found in a urine drug testing laboratory, with some exceptions.

We do prepare our standards as working stock solutions and we then spike them into the same matrix as the samples. This is typically intact hair for our screening assays and pooled negative digest for our confirmation assays. We use our standards to calibrate the assays.

We prepare the controls also in working stock

solutions and, again, in the same matrix as the samples. And we prepare the controls from a different source or lot number than the standards and we use the stem to verify the assays.

For our negative hair, we do collect hair of any type, of any color, ethnicity, background from a purportedly drug-free population. Typically, we use laboratory employees who are subject to random drug testing. We collect the hair in such a manner as to exclude contamination. That is we do not allow the hair to fall on the floor.

We cut the hair into small segments and mix it thoroughly and then weight aliquots for testing and we store it at room temperature.

To prepare the pooled negative digest, we weigh

the negative hair and then we subject it to the same procedures as we do our samples, in that we do an enzymatic digestion. We do a centrifuge to remove the melanin portion of this sample and decant it to remove aliquots for testing, and we store this frozen.

To certify that either the negative hair or the pooled negative digest, we do subject it to both the screening assay and the confirmation assay, the GC/MS of the GC/MS/MS and we should have no signal in any of the assays.

standards in methanol. We prepare the working stock solutions in buffer. We certify them doing RA and crossovers and GC/MS certification, and we recertify every six months.

For the controls, we also start with primary

standards and we prepare them independently of the standards. Again, we certify by RA crossovers and GC/MS certification and recertify.

We use our RIA standards to prepare a standard

calibration curve. Our controls in RIA are prepared at or near the cut-off. One of the things that i unique to hair is that we do have a predigestion spike which we would term as the recovery and we have a post-digestion spike which we call the assay. This is to control the effect of the digestion on the analytes.

We do track our controls with Levy-Jennings

charts.

Our GC/MS standards in controls, again, we start with purchase reference standards. We use different sources and lot numbers of the standards and controls and we prepare these in methanol for the final volume of 100 microliters. We certify the GC/MS standards by GC/MS or GC/MS/MS. We

certify them typically against previously-certified standards and we recertify every six months.

For GC/MS, our standards are -- contain for the cocaine analysis, we include cocaine benzoylecgonine and cocaethylene all at five nanograms, methamphetamine and amphetamine at five and one nanogram respectively, THC acid at .5 picograms, the opiates, morphine, codeine, and 6-acetylmorphine at five nanograms, and for PCP we still use the standard curve, and this is two to 20 nanograms.

For the controls, we use a high and low control. The cocaine, BENC, the low at two and the high at 10, for meth it is two and should be 10 -- or, excuse me, it is five and 10 -- oh, that is a mistake. Oh, I see. The meth is two and 10, and the amp is .5 and five. The codeine, morphine, and 6-acetylmorphine are two and 10. The other are underneath there.

We do, again, track our standards and controls in GC/MS on Levy-Jennings. This is just a composite slide. I think that you saw some of these with Dr. Cairns' presentation yesterday. I will just run through these rather quickly. Here is the benzoylecgonine, and the cocaethylene, and here is our THC acid, the .5 up to 10 picograms.

We do have matrix effects with hair analysis. The exogenous effects that we concern ourselves with are things like preparations being put on their hair, dirt, and smoke.

We usually can remove this by using a dry isopropanol wash prior to our normal wash buffer procedures.

For endogenous effects, this is more important to the RIA assay than to GC/MS. We do minimize those using an optimal digestion concentration of 10 milligrams a hair per one mL of digest.

For the blind quality controls, we used a

previously certified negative hair as a blind negative. For positive controls, we do purchase hair from rehabilitation centers or we can prepare a pool of previously certified, previously positive samples after they have been kept in storage for the appropriate length of time.

One of the problems with using user hair is that some of the times the drug concentrations can be quite high.

For quantitative assays, that can result in an excessive amount of time spent on the positive EQCs. Ideally, we want the positive EQCs to be within the linearity of the assays without dilution and exhibit good reproducibility, remembering that these samples are heterogenous.

To prepare them we cut them to approximately two centimeters in length. We mix them thoroughly. We obviously do not have a homogenous mixture, but we do have a pretty good mixture. We aliquot 10 times and then we determine the mean drug concentration by GC/MS.

This is the performance of some BQCs. I am just going to run through these quickly. This is a low-level cocaine with pretty good reproducibility. The CVs are a little -- about 20 to 30 percent for cocaine and BE respectively.

Here is another cocaine. I have been running this one about a year.

Here is a marijuana blind QC, and it is at four picograms with a CV of 12 percent. Here is the same one that I have been running for cocaine. It is also a marijuana positive at one picogram with a CV of 17 percent.

opportunity to use retest samples because the drugs are stable in hair over time. We retest samples from anywhere from a week to a year after they originally reported as positives, and they can be evaluated similarly to the QCs.

We have two types of retest samples. We can either retest the original sample or with hair we can recollect a second sample which we termed the safety net.

THC acid where we have the original versus the retest of the same sample, and we get quite good correlation. It is similar with cocaine. We see the same relationship.

The safety net of the newly-collected sample, if

it is used to support the original findings, needs to be collected as soon as possible after receiving the positive result.

We do a physical comparison of the two samples.

You would be amazed at how fast some people's hair can grow in two weeks. We compared those test subject's initials and we do some -- we section the safety net if there has been more than 16 days between the two collections just to try to get the similar timeframe.

This is some data showing the original testing results versus the safety net. We do not get quite the same correlation because, again, sometimes there are haircuts and their is time between the two collections. So, we are looking at somewhat of a different timeframe.

Similarly, we see the same pattern with cocaine.

I am going to conclude here with some typical

results that we see with some safety net results. These are two samples that were collected approximately two weeks apart. This person did get a haircut. When we retested the sample, we got the same result, but the safety net shows a decrease due to probably cutting off the positive portion of the hair.

Here is another THC result with almost a month between the resection of the sample. We get almost the same results of the sections.

Here are some cocaine results. I have got quite a few of these slides. I am just going to go through them quickly. One of the things I would like you to notice is that we do get very good reproducibility not only on the RIA levels, but the wash kinetics are almost identical from the first sample to the second sample. When there are time differences between the two collections and we do section it, we get pretty much the same result. In this case, there was almost a month between so we did section the sample.

Here is one where we did not section -- although this person seems to have very short hair, and there is some variability in the hair length of the sample. But, again, quite reproducible results with very similar wash kinetics.

between. We would have normally tested the safety net at 1.3 to 5.2 centimeters, and you can see that there is a

the same kind of reproducibility. This is very typical again. We have done about 2,000 safety nets while I have been at Psychomedics. When those are done for the purpose of verifying the original results, we do get quite reproducible results. Here is just another methamphetamine.

Again, the wash kinetics are almost identical.

We do also have, for some of -- we do have a blind program -- or a double-blind program where some of our clients can submit samples through William Walsh and Associates, and they can submit samples to -- he submitted samples to them and they submit it to us, and is monitoring us as sort of an external/internal QA program.

Thank you.

[Applause.]

DR. BAYLOR: The next speaker is Dr. Tamara

Nichols-St. Claire. Dr. St. Claire oversees all aspects of the PharmChem Laboratories Quality Control Department and reports to the Director of Laboratory Operations. Dr. St. Claire joined Pharmchem in 1995, and supervises a staff of seven individuals and validates all analytical standards and controls for the laboratories.

Dr. St. Claire received a Ph.D. in bio-inorganic chemistry from the University of California-Davis, and a bachelor's of science degree in biochemistry from San Francisco State University.

Prior to joining Pharmchem, Dr. St. Claire was an instructor at Cunsumne's River College, and a chemist at the University of California, San Francisco River Research Laboratory. Dr. St. Claire is a member of the American Chemical Society, and the Association for Women in Science.

Dr. St. Claire.

Agenda Item: Sweat Testing

DR. ST. CLAIRE: The quality control scheme and

the evaluation of the quality control materials in the qualitative and the confirmation assays was developed based on widely-approved set-ups using urine testing. Of course, the control levels have been altered appropriately to concentrations associated with the lower cut-offs in sweat testing.

The purpose of the QC program is to ensure consistent and stable performance of assays, or at least the results are only recorded from analytical batches showing acceptable performance.

For screening assays, consistency is judged

against performance standards established in the FDA Review Process. The cut-off is determined as a concentration and a calibration range that gave the fewest false-positive and false-negatives resulting in the highest percentage of clinical sensitivity and specificity. So, once we determined the cut-off, we needed to determine the control levels.

The analytical sensitivity of the cut-off was assessed by analyzing in replicate type samples at varying concentrations. For example, I will talk about the amphetamine assay. Despite concentrations of 2.5, five, 10, 12.5, 17.5 and 25 nanogram per mL concentrations, which

corresponds with 25 percent, 50, 100, 125, 175 and 250 percent of the cut-off.

The percent positive response is graphed for each concentration. The results yielded the threshold graph shown here in this slide. We chose a concentration that yielded a 95 percent probability of screening positive for our high control and a concentration that yielded a 15 percent probability of screening negative for the level of our negative control. Similar analyses were conducted for each drug assay.

For confirmation assays, performance standards are established in assay validation studies conducted by the laboratory.

There are several points of parallelism in the

sweat testing to the urine testing. The first one I would like to talk about is the validation in the quality-control materials.

Calibrators, and controls, and blinds are

evaluated in essentially the same manner as the QCs in the urine testing. Purchased ELISA calibrators are verified with the receipt that each new lot by GC/MS and they must quantify within 10 percent of the target concentration to be released for use.

The controls and blinds used in the ELISA assay

are validated by GC/MS and ELISA and GC/MS controls are

evaluated by GC/MS.

This is the cover sheet for the GC/MS calibrator evaluation. The lot under evaluation is compared to the previous calibrator and the reference material. Typically, the calibrator and the reference material are prepared from drugs purchased from two different vendors. If this is not possible, the stock solutions are prepared from two independent weigh-ins.

The samples are run 10 times and the mean must be within 10 percent of the target concentration and within 10 percent of the expected ratio, compared to the reference in the previous calibrator.

The quality-control scheme in the ELISA sweat testing is essentially the same as in the urine testing. In ELISA there is a drug-free, and negative control, that is a control that is below the cut-off concentration, and a positive control, one that is above the cut-off concentration. For the controls to meet the criteria for acceptability, the negative must have an absorbance that is greater than the cut-off and the positive must have an absorbent that is less than the cut-off. On the calibrators which are run in triplicate, we must have a CD of less than 20 percent.

The GC/MS batches used for confirmation in the

sweat testing also mimic urine confirmation batches. These controls must constitute 10 percent of the samples in the batch, and they must qualify as a target drug according to retention times and ion ratio criteria, and must quantitate within 25 percent of the target concentration for the client's specimen data to be reportable.

As you can see in the sweat confirmation batches, there is a calibrator, a non-extracted calibrator on the control, an open control, a carry-over check, and a blank after the carry-over check, as well as an additional QC.

has two blind quality control samples per 40 samples. In

the parallel to urine testing, if the blind fails, the batch fails.

In terms of trend analysis, GC/MS controls are

plotted per batch on the Levy-Jennings plot and reviewed on a weekly basis to evaluate the QC's performance and to allow for us to take corrective actions.

In terms of matrix effects, matrix effects are universal in all drug-testing assays. In fact, we can think of adulteration as matrix effects to such an extreme degree where the tests become susceptible to conditions of the specimen.

With automated analyzers, we have been able to

attenuate the effects of temperature, pH, read times, optical conditions, and other things to a degree that we could almost disregard them. This could be illustrated in the immunoassay techniques used in onsite testing devices. I think that Dr. Salamone addressed this just briefly yesterday. We have data that show temperature pH and specific gravity that show remarkable effects when removed from the analyzers.

This is a summary of the data collected with the cocaine device at four different pHs. As you can see, cocaine has an acceptable pH range of greater than two and less than 10 by comparing the expected responses at 450 nanograms per mL. You may be aware that a sample with a pH of 10 at a 450 nanogram per mL concentration would screen positive on an emit analyzer with no problem.

The effect of pH on the THC devices is even more dramatic. As you can see on this slide, they perform up to grades at a clinically-normal pH of five. An acceptable pH

range is much narrower in comparison to the cocaine.

After exploring pH, we thought it would be very interesting to look at the effects of temperature. There are three distinct temperatures you can imagine using the testing device sample. One is the void temperature; the second is the room temperature, that is, if you allowed the sample just to sit for a while before you test it; and the third is the refrigeration temperature, that is, if you store the sample to test it for later. We saw dramatic effects.

The cocaine devices provided surprising results at the void temperature compared to the room temperature or that of four degrees.

The THC devices gave the exact opposite results. Refrigerated samples gave false negatives at 75 nanograms per mL.

Let's return to the flat testing. As you can see, nature's effects are universal. So ELISA is really not immune to these effects. No pun intended. In fact, we use methanolic extraction solvents which can be quite severe on the immunoassay. We toyed with several extraction buffers.

In terms of extraction capabilities, we found that a 75 percent methanol, 25 percent .2 molaracetate provided to be very effective. But you can imagine how methanal could denature the antibody as well as strip it from the well.

assays for THC. In clinical trials, we have observed that the THC really presents the most effected assay in terms of matrix effects. However, there appears to be some component in sweat that actually interferes with the binding of the drug in the assay, and that gives us a negative bias.

The technique we have incorporated to control for this phenomenon in sweat is by spiking our calibrators onto warm patches. As you can see the calibration curve shifts dramatically. In fact, if you look at the range between 1.5 and 10 nanograms per mL without the shift, we would actually be losing quite a few positives.

In the developmental phase, we worked with RIAs as screening assay. Due to the non-specific binding in this technique, we showed a positive bias. To address this issue, we actually offset the calibration level to compensation for the high bias.

In Dr. Armbruster's presentation, he showed off

many of our GC/MS chromatograms. With respect to the GC/MS assay, sweat has actually proven a cleaner matrix and the chromatograms are quite clean, and have afforded us the ability to achieve reliable cut-offs at lower levels required for the sweat testing.

In terms of sample subtypes, there is a lot of

data that has shown that hair color and ethnicity play a role in the matrix effects with respect to hair testing, but we really have not explored that arena.

Thank you.

[Applause.]

DR. BAYLOR: The next speaker is Dr. Paula S.

Childs. Dr. Childs has worked in the field of toxicology and clinical chemistry for over 20 years. She earned a bachelor's of science degree in chemistry from Nazareth College in Rochester, New York in 1970. She then studied at Tufts University and earned a doctoral degree in chemistry in 1974. Following completion of the doctorate, she completed an additional two years of postdoctoral work at the Medical College of Virginia in clinical chemistry.

At the completion of additional studies in

forensic toxicology, Dr. Childs earned recognition as a Diplomate of the American Board of Forensic Toxicology.

Among her employment experiences, she includes a three-year term with the Georgia Bureau of Investigation working in the area of medical examiner, toxicology, and the Driving Under the Influence of Alcohol and Drugs Program.

She moved to the Compuchem Laboratory in 1986 to direct the laboratory's quality control program. She has been interactive in the development of various components of workplace testing programs.

In 1988, Dr. Childs was invited to assist the

Department of Health and Human Services in developing the standards for laboratory tests for the Drug-free Workplace Program mandated by President Reagan in 1988.

She was a member of the team of experts who developed criteria for the Laboratory Inspection Program, which is still in use today.

In addition to overseeing laboratory work, Dr.

Childs is a member of several professional organizations, including the American Academy of Forensic Sciences, the Society of Forensic Toxicologists, and the American Association for Clinical Chemistry. She has served on a variety of committees for these organizations.

Dr. Childs is recognized as an expert in her

profession. She has been qualified as an expert witness in forensic toxicology in more than 100 cases, including homicides, driving under the influence of alcohol and drugs, workplace arbitration, controlled substance abuse and military courts marshal. Her testimony has successfully supported law enforcement programs and workplace testing policies and procedures. Dr. Childs.

Agenda Item: Urine Testing

DR. CHILDS: Good morning. Thank you, Dr. Baylor.

I would like to extend a word of thanks to the Drug Testing

Advisory Board for the opportunity to present this information.

I would like to begin with a very basic definition of a quality-control program and that is to be sure that the test systems are producing the right results. Like Nike's, I like to do a play-off on Nike's "Just Do It," and just say let's just do it right.

I am going to focus a little bit, instead of just going through all of the regulatory issues related to definitions of quality control and talk a little bit about the focus of the urine drug testing because that has really become the basis and the foundation of what the other testing matrices and alternate matrices will actually be involved with as their technology is developed.

The QC materials for initial test are often very different from those in terms of target analyses and element composition from the confirmation testing. In addition, the QC requirements for initial test often require just a positive or a negative control response versus in confirmation testing where an actual quantitative value must be obtained.

In addition to that, during the initial testing, there is a requirement that at least one percent of the test batch should be a blind QC with a minimum of at least one and total complement of 10 percent total quality control; whereas, in confirmation testing, there is a quality control component, however, at this time, there is no requirement for a blind quality control to be part of the test batches.

obtained from confirmation testing with respect to 6-acetylmorphine and DL-isomers, as well as evaluation due to rate of internal standards that are part of a good quality-control program.

For definitions, I just want to mention just very

briefly that calibrators are solutions of known concentrations used to calibrate a procedure or to compare with a sample response to determine the positive or negative result; a control being defined as a sample used to monitor the status of an analysis to maintain its performance within desired limits; a standard being a reference material of known purity or a solution which contains reference material at a known concentration; and then QC sample is referring to any of those above materials used to evaluate an analytical procedure and whether that procedure is operating within pre-defined tolerance limits; of course, a sample being that of an unknown specimen or a donor specimen.

Analytical runs are defined to contain materials that include negative urine samples. These are materials that have been certified to contain no drug. And initial testing -- also positive controls being defined as those four to five with a drug or metabolite. Often this can be fortifying the negative urine samples to obtain the desired control concentration. And then also concentrations, positive controls have to also include concentrations at or near the threshold cut-off. Often that is defined as plus or minus 25 percent. I was thinking 20 percent with respect to the quantitative recoveries in the GC/MS controls.

Each initial test batch must contain negative samples, positive control four to five with drug, positive controls within 25 percent of the cut-off, and a sufficient number of calibrators to ensure and document the linearity of the assay with a method over time.

For confirmation testing, controls include: B-

negative -- negative urine samples, positive calibrators, and positive controls within 25 percent of the cut-off. In addition to that, if a test run includes analysis of limited detection or retest samples, then there is a requirement that there be a control at 40 percent of the cut-off.

For the validation and verification of QCM

materials, some very important criteria must be respected. One of those includes the traceability of the drug standard.

This would include documentation of the lot number and the manufacturer. Frequently, manufacturers provide source content. Other pertinent information would perhaps include a full scan of the product as it leaves their facility. They often include material safety data sheets that come along with the standard materials.

The laboratory then would verify with parallel

testing, full scan GC/MS and other analytical procedures that might include things like melting point HBLC and so on to perform independent verification from the manufacturer.

current lot numbers by parallel testing, also being able to use a reference, proficiency test, results from other proficiency programs to determine whether the materials being used in the laboratory are within acceptable tolerance limits.

When there are LOD and LOQ, and actually method validation studies, there are preparation of materials required in order to be able to evaluate things like the limit of detection and the limit of quantitation and upper limit of linearity for each assay.

The one thing that I wanted to mention here is

also the evaluation of test results by the certifying scientist. That is part of a good laboratory program and certainly is required under the federally-mandated guidelines. But the certifying scientists play a very critical role in being able to look at the test data, look at the quality control data, and make an assessment of acceptable quality control with respect to each individual

test result within a batch that they are certifying.

They are matrix effects. I wanted to talk about

the differences between urine or synthetic urine. Many laboratories like to use a synthetic urine matrix that may help eliminate some problems with contamination or different types of materials that are present in donor urines that would be collected. In collecting samples from donors within the laboratory, frequently, a preservative routinely sodium azide is added to the donor samples. It is important to understand that most donor samples coming into the laboratory as unknowns do not contain sodium azide. The typical concentration for sodium azide in the control materials being prepared is approximately one-tenth of one percent.

One of the comments I would like to add in the way that we prepare our pooled controls at our laboratory includes having various donors sign up as part of the donor program. We learned through years of doing this the wrong way first and then trying to get it the right way was that donors often use medications that are not controlled substances. When we were first doing amphetamine, you know, certifying samples to be negative for amphetamine, what we found was that a lot of people, especially during the fall and winter months are taking over-the-counter cold preparations which, by the way, do often interact with screening testing for amphetamines. So we found it necessary when we collected individual donor samples to test those samples before they were mixed in with the larger pool of other samples to verify that each of those individual donor samples were going to screen out the negative cut-off, rather than contributing some cross-reactive substances into the screening testing.

The same is true for GC/MS. Obviously, the GC/MS procedures should overcome many of those, but you want to make sure that there is no baseline effects. If you are going to be adding a fortifying negative urine with other drug, you want to make sure that it is acceptably at the negative level before adding anything else to that or pooling it together.

Just a word on synthetic matrices. I do not have

a lot of formulas and recipes. I have not been involved in actually preparing those. We know that some of the quality control manufacturers do use synthetic urine matrices for that. A couple of key things though. One is that synthetic urine is generally considered acceptable to use in a laboratory with the following conditions. One is that there be no matrix effect on the actual assay. That can be evaluated through some parallel testing with normal urine samples. The second is that it should be indistinguishable from human urine. The third thing is that it should be normal ranges for adulteration testing of the synthetic urine matrix issues. The fourth thing that is important to remember is that doing validation studies for the procedures should use human urine as the basis for the evaluation studies and not just a synthetic urine matrix because that can very profoundly affect the recovery and especially in GC/MS procedures.

I want to talk briefly about blind samples. There are basically two kinds of blind samples that laboratories blind. One are internal blinds, and these are generally defined as being blind to the laboratory's analysts. They can be positive or negative materials. On positives it is important to remember, to eliminate carry-over results, unless the procedure that you are using is set up to limit carry-over or to correct for carryover.

A lot of times laboratories want to site blinds to

just make sure that they are really going to be positive, so they put very large amounts of drug in the samples. That is, you know, just something to be aware of when preparing blind pools.

Preparation can be to fortify certified negative urine or to purchase materials from outside vendors. It is important to keep the lot numbers of the reagents and do the validations before they are actually introduced into the laboratory testing process.

In terms of introducing blind controls, they can

be introduced in coming into the laboratory, along with regular specimens, through the courier systems that a laboratory uses. In that case, they would be coming in in a specimen bottle probably with a chain of custody form, or the laboratory can introduce them into the system during the receiving or aliquoting process.

In terms of external blinds, many laboratories

like to have a feel for how other types of testing are happening besides just the analytical components, and external blinds can provide a very valuable source for looking at other administrative issues with regard to laboratory testing.

They should simulate real samples procedurally.

The lab then can track and monitor administrative things like turnaround time, follow-up with documentation errors, and other things that are surrounding the mandated programs with respect to chain of custody, and the turnaround time in certifying scientist-related issues.

Generally, the external blinds are not evaluated

in real-time along with samples, and the sample results are not held up if an external blind fails its criteria. Mostly that is because most of the people in the laboratory, including perhaps the certifying scientists, are not aware that these are external blind samples. The laboratory director in some audit fashion, would then review the results after the fact and make a determination of corrective action that might be required following a blind QC result. Generally, they are carried through the entire process that includes the initial and confirmation testing, as well as the reporting.

Trend analysis. We saw some very good slides earlier with different types of presentations. Generally, smaller laboratories tend to capture data in a manual manner, and may be able to plot it on a sample by sample or batch by batch basis. Again, it requires review, determination of means, acceptance criteria. With screening data, often the expected result of a quality control is either positive or negative. From a GC/MS perspective, Levy-Jennings often can create a statistical means standard deviations.

Most laboratories use a plus or minus 20 percent

of acceptability from a target value for the GC/MS expected result. There are West Guard rules that are being used in laboratories. These sometimes provide challenges to the laboratories in terms of determining which points can be eliminated. Again, that should be very clearly laid out in the standard operating procedure.

I want to talk just briefly about corrective

actions because often quality control results and failures should result in some type of corrective action. These would be quality control results that do not meet acceptance criteria should be entered into some sort of a problem log to try to determine what went wrong. How did the problem get remediated? Was the batch recalibrated? Were the control materials replaced? Were there samples that were clearly negative released anyway? That documentation is

very important. And then the responsible person or laboratory director would analyze the situation and then would develop a program for corrective action, and that should be reviewed by the supervisory staff involved in the set up and evaluation of those programs.

I wanted to just touch briefly on adulterants. We have had some discussions with some of the groups here about different types of adulterant products. It is possible to introduce, through the blind QC programs, different types of adulterants. You can introduce problem samples, be able to substitute, send into the laboratory, for example, Mellow Yellow and see how the staff actually detect that that is Mellow Yellow. Does it look like really urine and so on. It is the same thing with bleach and other types of products that are routinely seen as adulterants.

Again, follow-up with analytical problems would be what happens when there is a failure to recover an internal standard. What process does the laboratory go to to try to overcome that problem? Is there a dilution that is performed? Are there other types of reagents added? And, again, verifying that that is covered in the standard operating procedure.

With dilution analysis, that is important to

prepare and validate QC materials to be able to provide information on the detection limits and to also verify the reporting requirements. Often, one task, for example, may be that a creatinine is performed and, if it is less than 20, then a reflex to additional testing might take place. Again, QC materials to verify that that reflexing taking place properly.

A couple of special considerations. Preparation

of QC materials. Sometimes they are prepared lyophilized and have to be reconstituted. It is important to verify that the volume for reconstitution is providing the correct concentration of the drug after the reconstitution.

and concentration, and expiration dates, stability and storage. Are the QC samples being frozen? If so, make sure that they are being thawed and thoroughly mixed before they are introduced into the process.

Last, I just want to also talk about this very briefly, quality assurance issues. Laboratories often set up internal quality assurance programs to audit such things as instrument validations, temperatures, training records, making sure that all of the systems are in compliance, not only for the laboratory standard operating procedure, but with the program guidance that the laboratory is operating under.

Often this results in corrective action and identifying missing or absent information and preparing the corrective action to prevent such recurrence from happening.

Thank you very much.

[Applause.]

DR. BAYLOR: The final speaker on this morning's panel is Dr. Richard Anderson. Dr. Anderson received a doctorate in physical chemistry from the University of California-Davis. Dr. Anderson has 16 years in the development and marketing positions with Miles Laboratories, Hypertech, Incorporated, and Biosite Diagnostics. He has nine years of experience in the development of abused drug immunoassay kits. Currently, Dr. Anderson is the Director of Customer Support with Biosite Diagnostics. Dr. Anderson.

DR. ANDERSON: Good morning. I feel a little bit like the congregation trying to give a sermon to the preacher here, as probably a very junior member in the

forensic toxicology area. My experience is primarily seat of the pants in the medial diagnostic arena. I have a great deal of experience, however, in making assays, if not necessarily using them.

May I have the house lights a little bit?

I am going to speak on internal quality control as it affects onsite tests. I am going to speak to that topic really as a maker of tests, but attempting to address the issue for users of the test really in the operative mode.

verification and evaluation of quality control samples. The first one is calibrators. Calibrators, for most onsite tests, really, the calibration is done at the manufacturing level. That is done by the mechanism that the manufacturer generally unitizes the tests prior to distribution to the users. As a consequence, users really do not have an opportunity to "calibrate" tests, unlike they do perhaps with instrumented methodologies for doing drug tests.

quality control samples. As far as quality-control samples go, external quality controls and known quality-control samples are very much like those used for the standard reference laboratory.

They are generally multi-constituent, in our case, usually liquid controls. They are generally acquired from commercial vendors, not generally manufactured by the user themselves. They are, in some cases, done in that fashion, however.

One slight different wrinkle behind offsite tests that is usually not associated with instrument-based systems is that due to their utilization, they frequently incorporate on board or internal -- so-called internal controls. These control zones are designed to try to help in monitoring aspects such as test procedure, reagent viability and sample integrity. They do not -- these internal test control zones do not function as supplanting external controls. They are designed exclusively as adjuncts, however, they may adjust the frequency with which external controls are used. Their utilization is a topic of much discussion but greater acceptance in the medical community at this point.

Negative control samples are like positive control samples, generally purchased from commercial vendors. They may either be synthetic or natural pools, pools, for instance, from the pediatric patient populations.

Some users of onsite tests may, in fact, product their own negative donor and perhaps produce negative urine pools. Again, as an earlier speaker suggested, you have to be a little bit careful about that in that the negatives need to be verified as truly negative of all drugs, since some over-the-counter and prescription pharmaceuticals can cause problems in the screening exercise.

Onsite tests have a greater -- generally greater exposure to matrix effects, either physiologically-derived or non-physiologically-derived, such as adulterants. The reason for that is has been previously mentioned. Onsite tests generally do not involve any dilution of the sample. The sample usually is neat. And, as a consequence, the test has to survive the onslaught of all objects or items that are within the urine specimen.

I have listed a few items that are usually of

greatest significance to onsite testing, such as urine pH, specific gravity, because we new the tests involve filter membranes. Particulate load can be a particular issue. Adulterants, of course, as with regular immunoassays run in a laboratory are an issue, and colorants.

In an onsite testing environment, some of these issues can be addressed by the manufacturer themselves within -- by additions of reagents, particularly large amounts of buffer, for instance, to control physiologic urine pHs.

Specific gravity can sometimes be addressed by either observation followed by centrifugation. Particulate load is generally addressed by the filter at either -- filtering the material out during the course of the assay or occasionally clogging and you will see it in the controls.

onsite tests, however, relative to laboratory tests is with regard to adulterants. And that is that the on board controls provide a measure -- not a full measure, but a measure of guard against adulterants because the onsite internal controls have the ability, in some cases, to detect the presence of adulterants. That does not relieve the testing community of the obligation of attempting to determine the presence of adulterants by visual or olfactory detection in the preliminary sample, but it does provide an additional level of protection.

Blind specimens for onsite tests are a little bit more problematic. They are problematic from the perspective that he usual utilization of onsite tests involve the person doing the sample collection being frequently the person who also does the sample testing. In the traditional laboratory environment, the person doing the sample collecting frequently is the one doing the blinding of the specimens unless they are purchased as sample blinds.

This, of course, is a bit of a problem because it requires quite a bit larger number of blinds to be presented if the samples are truly to be blinded to the tester. An issue that I really believe has come up over the last day or so which I think may address a bit of the issue with regard to blind specimens is perhaps retesting a certain fraction of what are effectively blinds, in other words, negative donor samples or samples that are screened negative at the initial onsite screening test, to be sent on to the laboratory for rescreening and verification as negative.

I think that this might address a couple of

issues. One would address it or help out on the issue of blinding. Two, it would address, I believe, what has been referred to as the Bubba Syndrome, which is the confrontational component.

If a certain fraction of screened negatives are

sent on to the laboratory for rescreening, the mere fact that a sample has been held back, if you will, for additional testing, does not imply that that sample is presumed positive.

In terms of fractions, the question got asked, I believe, yesterday, of what sort of numbers might you add or select for rescreening. I believe that the numbers that were quoted were some that ranged in the range of five to 10 percent. I think it would probably be reasonable that the number of negatives is something of the order of the fraction -- demographic fraction of positives. Positive demographics, of course, tend to run in the range of one to five percent. So, something in the order of the mid-five range would probably be an appropriate number.

I would like to address a little bit about how one might select appropriate concentration for a quality-control sample in the area for onsite tests. One would have to remember that most onsite tests, in fact, all that I am aware of, involved a visual detection of the endpoint. They are not quantitative or numerical in the sense of producing an answer. They produce a yes/no result in that either

color indicates present of sample or it may indicate absence of sample. But the change from no signal to signal indicates the difference between positive and negative samples.

Nevertheless, the tests in an internal sense have

to act in an analytical relationship to concentrations of the samples.

On this graph, what I have attempted to represent

is interpretation of what a truly external quality control program burden places on tests, all tests, including onsite tests, for verification of positives or negatives. The red curve, which shows up only really on the left-hand side, represents a simple galcion where we have arbitrary spreads on the basis of standard deviations. The yellow line which has an abrupt transition of about 1.3 SD, represents -- the area under that yellow curve represents 90 percent of the total curve under the red area.

If you refer back to what the requirements are for any laboratory with regard to performance on proficiency testing, what is required is that 90 percent of all samples, at approximately 20 percent above the cut-off for that assay, must be judged positive and, consequently, approximately 10 percent are allowed to be judged negative.

transition is on the yellow curve to the positive/negative transition, in other words, the cut-off for the assay, again, you get 90 percent positives onto the right and 10 percent to the left.

If you calculate what that means in terms of an analytical CV, that allows an assay to have an analytical CV of up to approximately 15 percent and still provide results which would match this external QC requirement.

When you translate what that would necessarily

would allow for onsite tests or any tests for that matter, it says that, if you had an assay whose CV was on the order of 15 percent and you wished to have a hundred percent of results be positive, which means that you need to be at about three SD positive or to the cut-off, that says that the center point for the QC sample can be as large as approximately 145 percent of 45 percent greater than the cut-off. What we propose to you is that, for onsite tests, for -- eventually for the training purposes, it may need to be as large as that in some cases.

This is not really that out of sorts with some

regular manufacturers for laboratory-based tests in which the case -- after having looked at some package inserts, you find that the "near cut-off concentration" or positive controls are, in many cases, as large as 50 or 60 percent greater than the cut-off.

We can address the issue of trend analysis.

Because onsite tests generally have either a yes or a no final answer, and you are looking for effectively a change of state, not a trend in the general trending sense of a Levy-Jennings plot, you have to structure your sample in such a way that you are going to get all positives for samples which are supposed to be positive and not some fraction of them to be negative, which, of course, can be tolerated in a numerically-based assay when you are evaluating the trending in the context of the numerical result.

For corrective action, I believe it was stated

yesterday that -- I believe it was Dr. Armbruster who had mentioned this -- excuse me, it was the military individual -- that corrected action for offsite tests usually focuses on one of three areas: Either assay procedure, which generally because the procedures are relatively simplistic

in the number of steps that the operator needs to perform generally boil down to a re-examination of operator training; reagent replacement, either due to a single device failures, which sometimes are hopefully tracked by the internal on board controls; or perhaps full reagent lot replacement, just as happens with laboratory or instrument

based systems; finally, and the one I have actually found to be the most frequent, is that the controls themselves have, in fact, failed because they are probably one of the weakest links in the entire system, given that onsite tests

generally involve the utilization of liquid controls which have relatively limited shelf lives once the vial has been opened.

In summary, I would just like to say that the

internal quality control systems for onsite tests really mimic what would occur with laboratory-based instrument systems and for the simple fact that the onsite tests systems are simply the instruments compressed down into very small footprints effectively. As a consequence, the methods and issues that we need to address are really one in the same.

Thank you.

[Applause.]

Agenda Item: Questions and Answer

DR. BAYLOR: That concludes the formal presentations for the internal quality control program session. I will now open the session to questions from the presenters and the DTAB.

DR. KWONG: Good morning. I have a couple of questions for Dr. Gordon. If I understand correctly, your control material -- what you did was really spiking the digestion mixture.

MS. GORDON: We spike both digest and intact hair, yes.

DR. KWONG: But the data you show on the slide was spiking into the digestion mixture.

MS. GORDON: The GC/MS data is spiked digestion, yes.

DR. KWONG: So, you know, if you are not spiking

to hair, then what you are really controlling is controlling the recovery or the extraction from the digestion mixture, not from hair.

If, as you said, you have spiked some drug

directly into hair, did you show that data there?

MS. GORDON: I had one Levy-Jennings chart which

was spiked prior to digestion. That was early on. That was the RA results. We do also carry some of our pre-spiked samples into GC/MS.

DR. KWONG: The question I have really is, if you prepare your standards similarly, by spiking into the matrix or into hair, how do you assess the accuracy of the assay?

the blinds as well, which are -- we have the -- we do have pre-digested -- I mean, pre-spiked samples, and we do have -

DR. KWONG: The pre-spiked sample, how is it

spiked?

MS. GORDON: We spike prior to digestion. We do have some intact hair that is spiked prior to digestion.

DR. KWONG: So the drug is spiked into hair?

MS. GORDON: Yes. But most of our standards and controls are done for GC/MS after digestion into pool digest.

DR. KWONG: Okay. The only question I have is your double-blind.

MS. GORDON: Uh-huh.

DR. KWONG: What is assay line? What would be your target result for those?

MS. GORDON: For the double-blinds?

DR. KWONG: Right.

MS. GORDON: The ones that are submitted through our clients?

DR. KWONG: Right.

MS. GORDON: I do not know exact targets of those.

Sorry.

DR. KWONG: No, no. But my question is what would be your target result, just positive and positive?

MS. GORDON: I am sorry?

DR. KWONG: Yes. How would you assess the result of the double-blind?

MS. GORDON: If -- if they are --

DR. KWONG: What will be -- if it is positive above the cut-off, that will pass? If they were positive, they would be correctly identified. If they were negative blinds that were correctly identified as negative, they would pass, yes.

DR. KWONG: All right. Thank you.

DR. WILKINS: I have several questions. I hope

that you will bear with me. The first question I would like to address to Dr. Peat, Ms. Gordon, Dr. St. Claire. Dr. Childs, it may or may not apply to the urine testing program. So, if it is not applicable, yell.

As has been pointed out in the presentations, one

of the primary goals of a quality-control program is to ensure reliable and quantitatively accurate results. Certainly that has been a focus of the urine drug testing program for quite some time.

In alternative matrices, our speakers have also pointed out that, unlike traditional urine assays, we need to measure parent drug as well as metabolite. That parent drug is very important for alternative matrix analysis. Therefore, I wondered if each of you could address the following question. I think this is important for reporting above cut-off concentrations. That is how do you control for conversion of heroin to 6-MAM morphine or cocaine to BE in each of your analytical processes? I think that is particularly important for the confirmation procedures.

DR. PEAT: Well, there are obviously standard

analytical procedures for monitoring heroin in the presence of MAM monitoring, benzoylecgonine in the presence of cocaine. It is not unusual to do those two drugs simultaneously in a number of biological fluids. I would not see it as a problem.

DR. WILKINS: I just wondered if, in routine practice, do you have a control to monitor for conversion so that you know the number that you get for heroin or 6-MAM morphine that you might be reporting. We have not gotten to the reporting section yet, so this might be a premature question -- that that number that you report is quantitatively accurate. If you do not report quantitative results, then this may not be a critical thing. But it could be if your cut-off is defined by a certain concentration of heroin or a certain concentration of 6-MAM.

That is what I am trying to get at.

DR. PEAT: Well, right now, we just confirm the presence of benzoylecgonine in saliva, so it is not a practical issue at this point for us.

DR. WILKINS: Okay.

MS. GORDON: We do include a hydrolysis control in our sample, so we do measure the amount of hydrolysis in cocaine to benzoylecgonine.

DR. WILKINS: Does that control go through the

digestion procedure for confirmation assays?

MS. GORDON: We do have a control that goes -- is, again, we do a pre-digestion and a post-digestion pool.

DR. WILKINS: But, as for RIA?

MS. GORDON: Also for GC/MS. Remember, we do have

a hydrolysis control where we do measure the amount of hydrolysis that takes place during the digestion. So we have a spike three digest -- we know what we put in there. We put just cocaine, and we do measure the amount of cocaine converted.

DR. WILKINS: Maybe we will come back to that. Because I think that I have a misconception about the controls and standards that go through the digestion.

MS. GORDON: I did not mention it in my talk.

DR. WILKINS: Okay. So, routinely in GC/MS confirmations, calibrators and controls, the ones that you had spiked on the slide like, for example, cocaine BE, cocaethylene I think it was, five nanograms, five nanograms, five nanograms. Those types of calibrators and controls are spiked to a liquid digest?

MS. GORDON: Yes.

DR. WILKINS: Okay. What additional controls do

you have?

MS. GORDON: We have a hundred nanogram cocaine control that we do measure, which is cocaine spiked, and we do measure the amount of hydrolysis.

DR. WILKINS: Okay. Thanks.

DR. ST. CLAIRE: And I am going to defer that to

Bob Fogerson. I think he is going to address that in his talk.

DR. WILKINS: Okay. Thank you.

My next question is, again, for each of the

alternative matrix individuals, and I think this applies also to you, Dr. Anderson, is do you find that there is a large variability in signal response in drug-free matrix from different individuals? That is, if you were to compare saliva, or urine, or sweat, or hair from a large number of individuals, do you find that there is a lot of variability on a blank response? If so, how does this affect your LOD/LOQ and do you routinely report down in that range? What I am trying to get at is how low do you go typically in these assays? Because I think each of the matrixes are very different in what I might expect from a blank response. How do you control for that on a daily basis?

DR. PEAT: I can speak to benzoylecgonine in

saliva. There is variation in the saliva specimens, there is no question, from individual to individual. When we do an LOD/LOQ determination with several different specimens, we get values in the range of one to two nanograms per milliliter of fluid and our cut-off is 10-fold higher than that at this point.

DR. WILKINS: Okay.

DR. PEAT: The difficulty -- there is more

chemistry involved in the extraction procedure than there is in urine. So you can get lower LODs and LOQs simply by improving the chemistry of the extraction.

DR. WILKINS: Okay. Thanks.

MS. GORDON: We also see more of an effect in the screening assay than we do in the GC/MS because we can do more clean-up in the GC/MS. But, yes, we do have a -- try to separate -- have a fairly good separation between where our cut-off is in the RA assay, as opposed to the matrix variation that we see among the negative samples.

DR. WILKINS: Okay.

DR. ST. CLAIRE: Similarly, I think that our screening assay is more effected. That is why we have not

been able to typically go with the 25 percent above and below cut-off. Our GC mass assay is quite clean and we do not see as much of an effect.

DR. WILKINS: I wondered if each of the speakers and, again, I apologize, this does not apply to Dr. Childs or Dr. Anderson, but, if you could clarify for me again the typical precision or percent CVs that you see with fortified standards versus the typical percent CVs that you see in real matrix analyzed samples in general. Mike probably will only comment on DE. What has your experience been? I am just trying to get a flavor for, if I were to look at data as I was looking at whether it is a packet or the literature, what would I expect to find?

DR. PEAT: I think you would expect to find very similar CVs to what you see in urine. I think, with the use of the improved chemistry and due to rated internal standards, and care and attention to the extraction, you are going to get very similar CVs.

MS. GORDON: I think that we have the same experience in that. Particularly in the GC/MS we have quite good CVs. In our blind samples, our CVs are quite a bit larger, but that is because we have a more heterogeneous population. I mean, the mixture is not a homogeneous solution that you would have in urine.

DR. WILKINS: I think it is probably my fault. I can tell that the way the questions are going that I was not specific enough. For example, would you expect that your typical CVs and fortified standards are less than 10 percent and your typical CVs in a true matrix sample are the same. I am looking at approximate percent CV.

DR. PEAT: Certainly I would, yes.

MS. GORDON: We see the same, similar CVs.

DR. WILKINS: Okay. So, yesterday -- I just want

to make sure I am clear -- yesterday we talked about with standards in one of the presentations that typical CVs were between five to six percent, which was what one of the speakers had mentioned. But, on your slides today, it looks like realistically, based on your blind QCs, they are probably about 30 percent?

MS. GORDON: The blinds are higher again.

DR. WILKINS: And those are true matrix samples?

MS. GORDON: Right.

DR. WILKINS: That is what I am just trying to

understand is what would I expect to see normally?

MS. GORDON: But I think that is also skewed

because we take, you know, several hundred milligrams of hair and mix it up so we do not have -- we may make the blinds. We get quite a bit of hair from these samples.

DR. WILKINS: So you would expect on an unknown patient specimen --

MS. GORDON: On an unknown, I think when you look

at the retest data, you can see that we get quite, you know, we get very reproducible results when you retest the same sample when it is the same timeframe. When you take, you know, longer hair and chop it, you do get a larger CV in that mixture than we would in the retest sample.

DR. WILKINS: Okay.

MS. GORDON: There is only so much you can take

off of someone's head.

DR. WILKINS: Okay. Thank you. Dr. St. Claire?

DR. ST. CLAIRE: I just wanted to clarify the

question. You are wondering in repeating tests on client samples what the CB is compared to standards?

DR. WILKINS: If there appears to be a difference.

Because there is not a lot of information about alternative matrices. Does there appear to be a difference between what

you see with fortified standards versus what you might see in true patient specimens? That is what I am trying to understand.

DR. ST. CLAIRE: I do not believe there is a huge difference.

DR. WILKINS: Okay. Again, related to QC acceptance ranges, Dr. Childs mentioned in her talk that typically with the urine drug testing labs it is plus or

minus 20 percent or two SDs is what they typically set as an acceptance range for quality-control performance. I wonder for each of the alternative matrix individuals if you might comment. Do you apply the same criteria, slightly modified criteria? What do you typically apply in the -- or what, in general, is typically applied in those settings not necessarily just in your lab, but what people in generally typically --

DR. PEAT: We are going to go in reverse order

this time.

DR. WILKINS: Okay.

[Laughter.]

DR. ST. CLAIRE: In terms of our GC/MS assays, the controls have to quantify within 20 percent of the target concentration, similar as urine.

In terms of the screening assay, the negative control has to be above the cut-off and the positive has to be below and the calibrators have to have a CB of 20 percent.

MS. GORDON: We have for the -- the GC/MS controls are plus or minus 25 percent because we have got it at lower trace levels. But in the RA we do have a plus or minus 2ST.

We do that at quantitative.

DR. PEAT: It is the same as we use for urine

which is screening controls, negatives -- negative, positive positives and the plus or minus 20 percent.

DR. WILKINS: I have got more. Does anybody else want to go first? Okay. I wondered again, Emory, I have another question for you, I am sorry. How do you -- do you have controls for the wash procedure that you use?

MS. GORDON: Yes.

DR. WILKINS: I wondered if you would just mind explaining it to me because I missed it someplace.

MS. GORDON: I did not include that.

DR. WILKINS: Okay.

MS. GORDON: It was only 15 minutes.

DR. WILKINS: Yes.

MS. GORDON: We do have controls. We do spike

because that is a lot easier to do because we spike wash buffer for the controls and we do have similar controls than we do in our assays. We also do analyze the washes of the blinds.

DR. WILKINS: So you have an acceptance criteria

or some type of criteria for evaluating this --

MS. GORDON: The quantitative acceptability --

DR. WILKINS: -- performance of the wash?

MS. GORDON: -- of the washes. But our wash

evaluation is based upon the ratio of the wash concentrations to what is in the digest. So measuring it, evaluating the quantification that we get from the wash data and then using it to play with.

DR. WILKINS: Okay. One more. Again, this may be covered in the reporting section later. I am not really sure what is going to be covered in which section yet. So, if this is a premature question, just say defer it to later.

comment. How do you routinely establish the linearity of the assay? Is it a pre-established value or do you assign

that with each assay based on calibrators? This is for confirmation now. And then, if you have a sample result that exceeds that limit, what happens? I may not have phrased that -- that might have been a confusing question. If you need to report a quantitative result and it exceeds the upper limit of the linearity of your assay, what do you do with that and is that a pre-established value or is it experimentally determined with each assay? This sounds like an inspection doesn't it? I apologize.

[Laughter.]

DR. PEAT: It is getting very familiar.

[Laughter.]

DR. PEAT: Obviously, the way that we do that is identical to the way we handle urine specimens which is every six months we determine linearity of the assay and then any value that is above that upper limit we report as greater than if requested to do so. But generally, in the insurance industry, whether it is testing or performing, they just care if it is positive.

DR. WILKINS: Positive or negative.

DR. PEAT: They do not care how much positive it

is.

DR. WILKINS: Okay. Thanks.

MS. GORDON: We do dilute our samples to get them

in our linear range.

DR. WILKINS: So you analyze once and then --

MS. GORDON: We do a linearity every six months on our instruments. We do that linearity.

DR. WILKINS: Is that RA or --

MS. GORDON: GC/MS.

DR. ST. CLAIRE: Similarly, we do linearity

studies every six months -- reported greater than linear range.

DR. WILKINS: Okay. Thank you very much.

DR. BAKES-MARTIN: I just wanted to ask a question of Dr. Anderson. You mentioned that some of these systems have on-board internal controls that will detect adulterants. Could you be a little bit more specific about what they are detecting and how they are doing it?

DR. ANDERSON: Well, as you know, most adulterants operate by interfering with the antibody/drug interaction. Some of the systems have part of the controls as an antibody not a used drug interaction. If the adulterant can effect that, of course, it does not do that in all cases, it will trigger that control. In some of the cases, what will happen is you will get the wrong response for that particular control whether that would be the control being on or the control being off, depending on the nature. Some adulterants interact with antibodies in the sense that they actually literally destroy the immunological reagent. Bleach usually operates by actually direct oxidation of the reagent. Again, what you will see or tend to see in many cases is that the adulterant will have destroyed the reagent associated with that control. That control will come up with the wrong result and direction of certain structure if you do not get the correct response for that control to view the assay as invalid. It does not make the sample analyzable by the method. It merely indicates that that sample is probably not going to be analyzable.

It is fair, however, to say that, just as with

anything else, the controls are not perfect in that regard and do depend, to a certain degree, on the degree to which the adulterant is present, whether there is a lot, a little or a medium amount.

One place that the onsite test though had a tendency to be a little bit more adulterant-resistant is

that the literature -- and I am talking about the lay literature, not the peer-reviewed literature now, the stuff we have got on the Internet for how to avoid drug tests, most of the literature is associated with methods of adulteration which are designed to interact with the generally enzymatic detection space of immunoassays. Most of the onsite tests do not use immuno -- enzymatic detection. Most of them utilize pre-colored particles. As a consequence, that component of the assay because it carries a color that is not directly attackable by things like salt, which is a common one to disrupt enzymes, to that extent, they tend to be somewhat less adulterant-sensitive.

can be designed that would be more effective for onsite -- effect in the disruption sense, for onsite tests. But the greater share of adulterants have been selected by the delay for the purposes of disrupting the enzyme component of the immunoassay which is generally the weakest link.

DR. HUESTIS: I would really like to compliment

you all, everyone on their presentations. They were really nice this morning. First, for onsite testing, since you have had the fewest questions. There is a tremendous controversy going on now that you alluded to with HCFA's decision about the number and amount of controls to run on the onsite tests. I wish you would comment on the amount and type of recommended external quality control -- so, meaning a sample that has a known value that is not included within your device. And how often do you think lots need to be tested, whether it be each lot, each shipment, or onetime per day? What type of recommendations?

DR. ANDERSON: Yes. Part of the FDA's submission for onsite tests, which go through the FDA clearance process, involves some commentary by the manufacturer as to what they think are -- or feel are at least the minimum required external -- additional external controls. In general, the recommendation boils down to still just like the current medical recommendation of some time period. Usually the manufacturer's recommendation is somewhat longer than the current CLIA-mandated timeframe of two per day primarily as a cost savings. Because there is -- someone said yesterday what we need is a 25 cent onsite test, and there are going to be 25 cent onsite tests to date that I am aware of (sic).

Personally, having worked in this area and, as Rosemary will say, battled it out perhaps with the CDC, and HCFA, and others, as we try to work through what does make sense in terms of QC, I think what people have found is that the standard recommendation of things like two a day, if you really evaluate it in the statistical validity sense, is much more of a feel-good and not really nearly as much of a catch for errors as one might like to believe.

When you look at MLSPEC testing requirements, for instance, for how many tests have to be done for the kinds of tests that onsite represent in a unitized sense, generally, you are -- and with knowledge of what the general true device failure rates are which are really quite small, which can be verified at the original manufacturer level, you have come to the disappointing result that, in order to really find the errors, you would nearly have to test an entire batch with QC reagents to find them. Of course, in an operational sense, as a -- for users, that really is just simply not a practical matter.

So, the usual recommendation for manufacturers for onsite tests is the frequencies are on the order of once per week, say once per month. Certainly, every kit shipment, because you need to check for issues that the manufacturer

cannot control, namely shipping issues. If -- because most of the users usually buy reagents at frequencies on the order of once a month, and that corresponds to once a shipment typically, those kinds of timeframes are the timeframes that people potentially use external controls something on the order of once a week, once a month, something like that.

DR. HUESTIS: And is that what the Joint

Commission accepted? I do not remember the exact--

DR. ANDERSON: Yes, it is.

DR. HUESTIS: -- the once per month?

DR. ANDERSON: At this point in time, the organizations that are utilizing that criteria are HCFA, under their clear regulations via the State Operators Guide, not actually directly under the CLIA regulations. The Joint Commission currently subscribes to that mechanism --

COLA, which is not usually in the hospital laboratory, but certainly affects the physician office laboratory environment.

Probably the largest organization that is

resisting this at this point is the College of American Pathologists. However, I will say that just this year, not in the drug testing arena at this point, but in a different arena, namely of electrolyte blood gasses, type devices, the ISTAT device, for instance, for that, electronic QC used in the point of care environment is now an accepted methodology even within the College of American Pathologists. For that matter, onsite drug tests which are used in the point of care, which is relatively uncommon at this point. In fact, under the point of care checklist, under the CAP, alternative QC -- speaking was about alternative samples today -- but alternative QC is a potentially viable route within the College of American Pathologists.

Again, I do not think any of the manufacturers of the onsite methods really would be willing to tell you that do not do extra QC at all. It is really a question of how do you split between internal methods and external methods to kind of get the biggest bang for the buck.

DR. HUESTIS: Okay. Thank you.

For the alternative specimen people again, please.

Could you describe for us the batch sizes that you are using? I know a lot of micro titer plates are 96 OLs. But, if you would tell us something about batch sizes. And then I think very importantly is the placement of the quality control within the batch. You all showed us very nice controls, but it makes a difference whether at the beginning of the batch or spread throughout the batch and what kind of CVs you see and how you control for drift across batches.

300 samples in the batch. And we do have sample controls throughout the batch.

In GC/MS there our batch size can vary up to about 100 samples in a batch and every approximately 10 to 15 samples is a blind -- I mean, is a control.

DR. HUESTIS: So, you have controls throughout the batch? In the screening across the 300 you do not have much problem with drift, with the RIA?

MS. GORDON: Not too much.

DR. PEAT: With the ELISA we have to control every 10 specimens. We do not see hardly any drift across that plate.

DR. ST. CLAIRE: In terms of the GC/MS batches, we have the controls at the beginning and the end of the batch.

In terms of the ELISA, there is -- we use 96, and we also put the controls at the beginning and the end of the batch.

DR. ALAN JONES: I am just going to go down the table. I will start over with you, Dr. Anderson, for a question. A concern that is often reflected about all of the whole QAQC is the maintenance of records. What types of things are being done in the onsite arena for the maintenance and storage of QAQC-type records?

DR. ANDERSON: Well, of course, in general, most onsite methods do not generate a paper trail in and of themselves. They are not like instruments with a piece of paper that gets spit out the front. There are a variety of mechanisms for generating the paper trail, generally requiring a transcription of the results to some kind of form. In some of the devices it is possible to xerox or photocopy the device to replicate the results. But, depending on the quality of the photocopies, that may or may not be satisfactory. But, normally, it is done by the mechanism of simply recording the visually-observed results on a standard form.

DR. ALAN JONES: Of course, as we get into litigation procedures and the like, certainly, my experience is more in the urine testing side of it. Many of these types of records are brought into the proceedings. Is it your experience that these are the types of records that are satisfying the queries from the arbitrator, or the courts, or anything?

DR. ANDERSON: Our experience is that that sort of paper trail is not an acceptable mechanism for generating results. It mean, it certainly has the downside that, of course, transcription errors can be introduced into the record, if you will. Most onsite methods, however, do not totally consume the sample. The sample is capable of a split. If there is a challenge to the onsite initial screening results -- and remember that the initial screening result, or the onsite tests are only an initial screening. They are not the sum total of the paper trail.

When the sample has gone on for, as is

recommended, for further confirmation at a certified facility, there is a secondary component to the paper trail, which, of course, supports the initial screening results. Should, for some reason, the initial screening result be -- either paper trail or the result itself be challenged, in general, there is a split sample that can be retested or rescreened and confirmed. So it provides enough safeguards I think that generally the results are accepted.

DR. ALAN JONES: But that would not be the case if it were a negative QC and did not go forward?

DR. ANDERSON: If the initial sample is a

negative, of course, the results are marked down in part of the paper trail. However, it is generally not going to be the case that negatives go forward for confirmation for onsite, just as negatives do not go on for confirmation at a standard reference lab.

DR. ALAN JONES: Sure.

DR. ANDERSON: However, I think, given some of the feelings with regard to that and the fact that I suspect that, if onsite tests are brought within the Federal workplace programs, that my expectation is actually that a certain fraction of the negatives probably will be included, if for no other reason than to sort of get the program off the ground. There will be a certain fraction of negatives, potentially false-negatives, or probably some fraction will eventually be false-negatives for that. That issue will be addressed.

DR. ALAN JONES: Thank you. Dr. St. Claire, I

have one question. Specifically, you showed us a slide talking -- well, I do not know where it was -- talking about

bias and the bias of the sample. Is that bias that you have seen uniform across samples or do you get a lot of variation in that bias? Can you expand on that a little bit?

DR. ST. CLAIRE: Are you talking about the THC

ELISA assay?

DR. ALAN JONES: Right. And the bias from the -- that you saw from using a patch, a pre-worn patch versus a non-worn patch in the preparation of that? You were talking about matrix bias?

DR. ST. CLAIRE: Right. That bias was only seen

in the THC assay. Actually, we did spike the calibrators onto worn patches with the other assays and did not see the similar effect. It was only after doing the THC assay.

DR. ALAN JONES: Only in the THC assay?

DR. ST. CLAIRE: Uh-huh.

DR. ALAN JONES: Is that -- do you have any feel

for that? Is that a nature of the antibody or is that -- what is the genesis of that? Do you have any feel for it?

before, I think there is just some non-specific binding that is going on.

DR. ALAN JONES: Okay.

DR. ST. CLAIRE: I have a question to Dr. Peat,

Ms. Gordon, and Dr. St. Claire. As your particular assay may be moved to another lab that does not have the experience that obviously the three of your facilities have in your given area, what kinds -- and this gets into a little bit more beyond QC, but we do not have a section on validation, so this is a little bit of a validation question I guess. What kinds of challenges would you recommend people consider as you might generate or build QC specimens for validation and looking at ruggedness of the assay? Obviously, you are going to look at the analytes toward which your particular assay is directed, as you have indicated. Are there other things that we should include in these challenges or anything, as another lab might pick up your assay without the experience that you have in your particular arena? Does that make sense?

DR. ST. CLAIRE: Dr. Peat, I would like to defer that to you.

[Laughter.]

DR. PEAT: I am sure you would.

DR. ALAN JONES: Dr. Baylor, would you like to

handle that question.

DR. BAYLOR: No, thank you.

DR. PEAT: The biggest challenge, I believe, that other labs are going to face is not necessarily validation protocols. I think they are generally similar to what you would do in the urine program. It may be that the interfering substances may be different.

I mean, for example, doing THC you would have to involve yourself in making sure that other lipids and fatty acids, et cetera, did not interfere with the assays. But I think that the biggest challenge that people are going to face is not in modifying the immunoassays or adapting the immunoassays. They are pretty accepted, commerciallyavailable procedures in both cases, FDA-approved.

I think that the biggest factors are, one,

carryover. When you look at alternative fluids, any hint of a carry-over could possibly lead to a positive because you are looking at much lower concentrations. So, carryover, in a sense, has a much greater potential when you are using pipetted diluters in 96 well formats. Then it might attach here on an instrument.

I think that the other challenge people will face

is the ability to confirm at very low concentrations with

the degree of confidence that these programs require. You know, as I mentioned in my slides, and as I have mentioned before, I think that trying to do sub-10 nanogram per mL determinations with mass selective detectors requires a bit of art. In some days, it probably requires you to kneel on the floor and bow towards HP headquarters.

[Laughter.]

DR. PEAT: That is the biggest challenge I feel

that people will have with these alternative technologies. It is not necessarily the validation protocols. They are pretty similar, but just being able to do these assays accurately and reliably day in and day out.

MS. GORDON: I am not sure I can add anything to that. I think I agree with it.

DR. ST. CLAIRE: I think that Dr. Armbruster

actually addressed this to some degree yesterday.

DR. PEAT: Right.

DR. ST. CLAIRE: I think we are still somewhat in the infancy in terms of our ELISA assay. I think it would be a wonderful challenge to get to another lab to really try to have controls at the levels that we are able to in the screening, in the urine screening assay, 25 percent above and below the cut-off.

DR. ALAN JONES: Thank you.

DR. CAPLAN: I just had one follow-up question for Dr. Anderson. You were asked earlier about, and maybe Sal might want to comment on this, about how the adulterants might be detected in the various onsite devices. Particularly, I think you made reference to the effects on the assay channel. Some of the devices like I think the one you have, does have a separate channel to validate the process and some other devices only have the assay channel.

Could you comment a little further on whether there is an advantage to a control internal channel for these processes or whether the assay structure themselves would -- how that might detect or deal with the adulterants?

DR. ANDERSON: Yes. I guess -- and this is an

issue that those of us who were in the onsite group, as a whole, sort of struggled with -- is that, unlike the other alternate matrices, there really is a single system that we are talking about. The onsite tests are, if you will, a collection of a variety of manufacturers with a variety of tests and the right formats. So the comments, as a consequence, are a little bit generalized in that regard.

systems have as many on board controls as some others do. I personally think that, essentially, the more the controls probably the better off you are because they provide -- the different controls can respond to things such as adulterants or other intrusions into the assay procedure in different manners and give you just that more opportunity to trap out errors.

It is true that some of the systems work better

with some of the different adulterants. I really have to comment more directly with our own manufactured device. We know that, for instance, some adulterants affect one of the two control zones more directly than they do the other of the two control zones.

Obviously, having only one of the other of the controls would have tended to make your more vulnerable to that specific adulterant. So, I guess, the only way I can answer the question is that the more safeguards you put in the system a little bit, the better off you are. That is really the reason for not taking external controls out of the greater QC environment, but taking advantage of the QC elements that internal controls do provide. I do not know

if that answers the question.

DR. CAPLAN: Okay. Well, I guess, the corollary

is do you feel a separate control zone in an onsite device would be an essential quality-control component?

DR. ANDERSON: I think it is a very valid component. I mean, you can ask the question in a slightly different fashion. If you think it is a really great idea, then why would you not have it in the instrumented systems?

The reason you do not have it in the instrumented system is because basically it is technically possible to do.

I mean, the difference about onsite devices from

say instrumented laboratory-based immunoassays is that it is really impossible to think of each surface area element on the membrane as a separate reaction area. It is really possible to do separate chemistry in the separate zones. So, if you have four, five, seven, eight, however many reaction zones, you really can do four, five, seven, or eight different chemistries. Historically, for the immunoassay systems, while it is possible to do homogeneous systems, such as the FPIA, it has not been possible generally to do multiple homogenous assays inside of this single bottle. People have tried it in some cases with some optical labels where you could measure it one way versus the other in the same tube. But, historically, trying to get that to work has been quite challenging or the assay manufacturers.

Again, I think, the more controls you can add, the more likely you are to be able to trap out errors that a less-sophisticated user might not be able to do.

But when it comes to adulterants and the like,

certainly the first wave of defense is not the assay. The first wave of defense is a good collection procedure and good collectors. They are the first ones to touch the sample and the first ones likely to notice that something is obviously awry with the sample, such as strange color, you know, funny temperature, strange odor.

immunochromatographic approaches, yes, it is easy to put a control zone in that. But, if you have different types of technology, sometimes it is impossible, just like it would be impossible to put an internal control on an instrumentbased test.

You are familiar with the on-track system. There

it is solid -- or it is reagents in the same way that you would use instrument-based reagents. There it would be nearly impossible to put an internal control in the same, while you are running the sample.

DR. WALSH: Just an FYI. I happened to be aware that Stewart Bogum has submitted a very elegant study where he has looked at all of the common adulterants on a number of the onsite, FDA-approved devices that have separate control zones. So you will have to go to Soft this fall to find out the results.

DR. CONE: I know time is getting long here so I will make my questions real quick. I had some onsite testing questions. There is some variability in reading. When we evaluated we had three readers read independently the results and got pretty good concordance usually, around 90 percent. But then there was 10 percent discordance. Do you see a need for more than one reader?

DR. ANDERSON: Our experience is that, in general, that is not a necessity. Certainly, as the concentration of the sample gets near the cut-off, there is variability in the interpretation of the result. That is a natural order event. I mean, in a certain sense, if we take an

instrumented system and look at it from sort of the same variability perspective, in the most careful definition of cut-off, which is that the sample essentially will produce 50 percent of the results greater than the numerical value of the cut-off, and 50 percent of the results numerically below the cut-off, in a sense, the instrument has essentially the same problem. And, from what I am aware of, there are no instrumented systems that have effectively a delta function as an error curve for their assay, that occurs over some, if you will, grey zone of interpretability for an instrument perspective.

Certainly, the same thing happens in an onsite system because it produces -- its color change varies to a certain degree over the range, over some concentration range. In most assays, and I am going to mean by that most drug assays, usually the samples do not present themselves very frequently actually at the cut-off. That is perhaps arguable in the case of THC where there are for reasonable numbers.

But most of the samples I think Dr. Peat said

earlier, they were either nothing, a lot, or a hell of a lot. And so, as a consequence, most of the results fall in the category of there is, depending on the system, either not much of a signal, a lot of the signal, or a bunch of the signals.

DR. CONE: But you will get samples of the cut-

off?

DR. ANDERSON: Oh, yes.

DR. CONE: A related question. You and others

have suggested sending out negatives for confirmation. That seems to me to be a good QC check. On the other hand, what do you recommend about if you get the negatives back with a positive result? Should that then be used as a -- this is more of an administrative issue, but it relates to QC. If you implement that procedure, then what do you do with the results?

DR. ANDERSON: Well, I think that is exactly what the issue is going to entail. I mean, certainly, when one goes to select any analytical method, be it drugs of abuse, electrolytes, what have you, one of the goals is to select the most bang for the buck, if you will, the best system you can possibly afford. Certainly, if your decision was that your onsite method was not providing an adequately small fraction of false-negatives, I would suspect that the user of that method would have serious doubts as to whether they would continue on utilizing that approach.

DR. CONE: But that is not answering my question.

What would you do with a negative result that was reported

back to you positive?

DR. ANDERSON: You are going -- assuming that that went through a confirmation at that point, you are going to have to assume that that result is positive and act accordingly.

DR. CONE: So, you would think it would be a valid result then and should be reportable --

DR. ANDERSON: Yes. I think that --

DR. CONE: -- even though you have a negative

result already established for it?

DR. ANDERSON: Well, let me answer that question

in a slightly different fashion. As you know, urinary drug tests are designed to detect urinary metabolites. That is why urinary drug tests are oriented for cocaine, for benzoylecgonine. We all know that quite well. There are many of the urinary metabolites of other drug classes, particularly things like benzodiazepines, where the compound is not benzodiazepine yet. There are quite a number of

immunoassays out there right now that are designed to detect pairing, which is not really the compound that you would usually see.

I guess I would ask the same question. What do

you do with a test that provides a large false-negative reading in that category. The answer is, if you find it is truly positive, you respond in that way.

DR. CONE: Do you have a recommendation for the local sites in terms of establishment of SOPs and related to the records question originally?

DR. ANDERSON: Our recommendations are essentially that those that fall into the regulatory environment, that the NCCLS protocols for setting up standard operating procedures are good guidelines to follow for facilities in setting up procedure reviews.

DR. CONE: Okay. One quick question for Marie. Have you guys -- do you have any way of producing quality control samples for hair that you are happy with in terms of other than using previously-known drug users' hair? Have you developed any positive control methods for making hair samples that simulate real positive hair samples?

MS. GORDON: I think that Dr. Baumgartner has

looked into this. We are not currently using anything like that at this time, but I believe that Dr. Baumgartner could better answer that question.

DR. KWONG: Ms. Gordon, I just wanted to clarify

my confusion about the QC data. I am sure that the audience will appreciate a clarification too.

You mentioned that some of your control data were from material that -- hair that the drug has been spiked into hair rather than into the digestion mixture. Is that the IA control recovery and IA control assay, those two slides?

MS. GORDON: The recovery is spiked into hair.

The assay is spiked after, so it is in to digest.

DR. KWONG: So, when you said spike into hair, you mean that you actually incubated a drug in?

MS. GORDON: No. We just added the drug to the intact hair.

DR. KWONG: How long did you leave it with the

hair?

MS. GORDON: Just a few minutes before the assay.

It is not subbed in the drug.

DR. KWONG: Okay. So, do you know --

MS. GORDON: And it is also very --

DR. KWONG: -- do you know the entry of the drug into hair in that short timeframe?

MS. GORDON: I beg your pardon?

DR. KWONG: Do you know if the drug will enter into hair in that short time incubation?

MS. GORDON: No, I cannot answer that. I do not know. I doubt it because it is a very short time period. We are not soaking the hair in the drug. It is really adding 100 or 200 microliters.

DR. KWONG: Okay. Thank you.

DR. BAYLOR: I would like to personally thank the members of the panel for their preparation of the presentations. We will now go on break.

The meeting will reconvene at 10:30 with the

reporting test results panel. Thank you.

[Applause.]

Agenda Item: Reporting Test Results Panel

DR. BUSH: We would like to begin this next panel.

This panel is entitled reporting test results. Okay. Well, we are going to get started. We have given everyone enough grace period here. Okay. Reporting test results

panel. Our mod