Could Paul Edwards, thrower banned for life, be telling the truth?

Ultraboy wrote:

I know what you’ve said NA. What you have not actually provided the document where KCL say that the machine was not calibrated. You have admitted that they have said it was calibrated and that the data for that calibration has now been destroyed. Given the age of the date the fact it’s no longer in existence is not surprising.

Your argument, and this has nothing whatever to do with disputing science, is that the test is invalidated because they didn’t calibrate it specifically for Paul’s test. If the code from the time says calibration is required for each test then you have a good case. If it doesn’t then you don’t. But you haven’t provided a copy of the code.

The bit you seem to have difficulty understanding is that saying it hadn’t been calibrated expressly to do Paul’s test is not saying that it was not calibrated correctly to do the test. It’s a critical distinction. To use an earlier analogy a speed gun isn’t required to be recalibrated for each and every vehicle that it is used on. So you could argue it wasn’t calibrated for each speed check. But that would not mean that it was not calibrated to the required standards.

Ultraboy wrote:

If you cannot tell us exactly what Beloff asked for then how can you say that it shows what was required?

You seem to be constantly moaning that it’s not possible to get information from so long ago yet also moaning that Time Barring is wrong.

Lets be clear, experts are usually in a case on both sides arguing different viewpoints. I’m not going to sit here and just blindly take what you say as fact when you can’t dven provide the code you think was broken.

As usual just more personal insults.

Hers a summary. You don’t have the code, you don’t know the actual standards. You can’t even provide information from the two hearings, it’s all conveniently lost.

What you have is some opinions not based on the standards from some experts. Professor Cowan is an expert, Beloff is an expert, the panel convicted your guy twice whilst being fully aware of the standards applied at the time.

You don’t want to answer awkward questions because you don’t have an answer so you resort to personal attack. Just a bunch of bully boys who don’t want to debate and clearly only cane to ketsrunto try to get Rojo and Wejo to help you.

Lean Cuisine wrote:

"Review of new analytical data withheld from Paul Edwards and the BAF/UKA disciplinary and appeals panels

Just a reminder as has not been posted for a while.

Simon Davis PhD

INTRODUCTION

Note to reader

To understand fully the issues relating to the analytical data which was withheld from Paul Edwards’ team, it is first necessary to understand the basic principles of the analytical technique used by the Drugs Control Centre to analyze Paul Edwards urine sample. The technique in question is known as Gas Chromatography Mass Spectrometry (GC-MS). It is a well-recognized method of identifying and quantifying steroids or steroid breakdown products (metabolites) in an athlete’s urine. If the reader of this report is not familiar with this technique, I suggest that he/she reads the brief introduction held in Appendices A and B.

Int.1. On the 09 of May 2002, Mr. Caborn, Minister for Sport, stated in the House of Commons in the Adjournment Debate which was introduced by Andrew Hunter MP that Paul Edwards had received all the data relating to his doping case. It has subsequently been discovered that this was not the case: a further 600 pages of undisclosed analytical data had been withheld by the Drugs Control Centre (DCC). Of course, the Minister did not knowingly mislead the House; however, the briefing which he received prior to the Adjournment Debate from UK Sport, his officials and perhaps other advisers clearly did not include the data which had been withheld by the DCC.

Int.2. This data had not only been withheld from the Minister and Paul Edwards’ team but also from each of the disciplinary and appeals panels before which the case had been brought. A life-ban was imposed on Paul Edwards without any awareness of the data which is discussed in this report.

Int.3. Paul Edward’s team has been aware of the existence of this data since the outset of the case against him. We have campaigned vigorously for its full disclosure. Obstacles have been put in the way of our obtaining. We have met with prevarication and delaying tactics.

Int.4. We requested in writing prior before the first hearing that we could have sight of all data. Our request was unsuccessful. Subsequently, we have submitted several requests for data under the Data Protection Act to the British Athletics Federation (BAF), UK Sport, UK Athletics (UKA) and the DCC. Despite its statutory obligation, the DCC refused to disclose any data to us. We therefore made a formal complaint to the Data Protection Agency, which sent a formal written warning to the DCC. The DCC ignored this warning and continued to withhold the relevant data. The DCC claimed that because the data was held on ‘continuous feed paper’ (i.e. all the sheets were joined together), it would require “disproportionate effort” to photocopy.

Int.5. In order to gain access to the data, two legal representatives and I used the Data Protection Act to gain access to the DCC laboratory in order to view the data which “required a disproportionate effort” to copy. Despite the paper being joined together, it took one person around thirty minutes to photocopy the entire bundle (significantly less time than it took to collect and record Paul Edward’s urine sample!). The effort invested by the DCC to prevent the release of this data over the previous five years had been significantly greater than the thirty minutes of “disproportionate effort” which would have been required to photocopy the bundle.

Int.6. On initially looking through the bundle, we noticed that it included a number of hand-written directions. A blue post-it tab on the top of the bundle was particularly interesting. It read, “Just the T/E sheet; don’t give attached data”. At this stage it is relevant to recall the International Olympic Committee (IOC) ruling:

“Copies of all analytical results shall be made available from the laboratory when requested by an appropriate authority” [Appendix d, sub-section 1.2 Olympic Movement Anti-Doping Code].

It appears that the DCC feels that it is not obliged to follow its own regulations or inform Ministers of State to the existence of data, if it does not wish to.

Int.7. A review of the data in the bundle quickly established that yet further data is still being withheld by the DCC. Crucially, this included the results from the standards and calibrations used to identify and quantify any Testosterone (T) and Epitestosterone (ET) in Paul Edward’s sample. Withholding this data makes it impossible to validate the DCC’s findings or identify any possible errors in the DCC’s results (as occurred in the DCC versus Mark Hylton). In addition, the results of Paul Edwards Luteinizing Hormone (LH) tests are also still being withheld, despite being used as part of the evidence of Mr. Edward’s guilt. Furthermore, the results summary form (tab 6) refers to a further two GC-MS analyses for which the data is again absent, whilst tab 1 of the bundle refers to a further GC-MS-MS analysis whose data is also missing!

Int.8. It is clear that crucial data which indisputably relates to the Edwards’ case is, to this day, being withheld by the DCC in contravention of the Data Protection Act. If the case against Paul Edwards is sound, one can conceive no reason why this data should not have been released immediately in order to validate the DCC’s finding

Review of new data evidence obtained from the DCC

Int.9. All points and comments below relate to the new evidence which was withheld from Paul Edwards’ team and from the relevant disciplinary and appeals panels.

Int.10. After the reviewing the new data from the DCC, a number of serious questions were raised over the validity of the analysis which had been made of Edwards’ sample [010830]. I address these in this paper in the order listed below:

Apparent use of incorrect ion to calculate Paul Edwards’ T/ET ratio.

Inconstancies in the volumes of urine used in the analysis of sample 010830.

Review of A sample analysis of urine 010830.

Poor identification of T and ET in the A sample.

Review of further analysis of urine 010830 (A and B).

Poor identification of T and ET in samples run between the 18th and 20th of September.

Lack of calibration data for all sample analysis.

Improper use of 20 to 1 standard.

Large tri-deuterated peak in Water Blank analysis.

1. Apparent use of incorrect ion to calculate Paul Edwards’ T/ET ratio

1.1. Within tab 6 of the bundle which we obtained from the DCC, a results summary/calculation table is presented. Although the DCC have refused to explain the method of calibration, there is clearly a gross error present within this sheet. The values used for the peak heights of T are labelled as having been measured using the ion 435 (m/z 435.3). This ion is in fact the tri-deuterated internal T standard (3dT) and not the T from Paul Edwards’ sample.

1.2. If the T of the athlete’s urine has been mistakenly confused with the internal standard, then clearly the reported T/ET ratio becomes meaningless. This rather basic error draws into question the validity of the whole test and also invites the question: what other errors have been made?

1.3. Professor Cowan has commented on this point. He describes it as a minor “transcription” error. However, despite four Data Protection searches, we have found no documentary evidence that this error had been either identified or rectified prior to the T/E ratios being released to the disciplinary panels. Two conclusions can be drawn: either that an error went uncorrected and all the reported T/E ratios are wrong; or that the error was corrected and, despite four Data Protection requests, documentary evidence which establishes that corrections were made have been withheld.

1.4. Paul Edward’s team has been unable to investigate this point further as the calibration methodology and data necessary to do so has been withheld by the DCC. If this is a minor transcription error, we are bewildered as to why the DCC refuses to release the data which will prove it so.

2. Inconsistencies in the volumes of urine used in analysing sample 010830

2.1. Professor Cowan describes the method for sample preparation in his letter of 2nd July 1997. He stated that each extraction required 4 ml of urine.

2.2. Within the bundle supplied by the DCC, results from fifty-one GC-MS analyses of sample 010830 were printed. The bundle also contained print-outs of a further three HPLC analysis (tab 29) and at least one GC-MS-MS analysis (tab 3). A further GC-MS-MS analysis is referred to in the event log (tab 1) but the relevant data is missing. Also, the results summary form (tab 6) refers to a further two GC-MS analyses for which the data is also missing.

2.3. If the DCC had followed its own written protocol and used 4 ml of urine for each extraction, over 200 ml of urine would have been required to perform fifty-one GC-MS analyses. Bear in mind that this excludes the HPLC, GC-MS-MS and missing GC-MS-MS and GC-MS analysis.

2.4. From the doping control booking form it is clear that only 170 ml of urine were provided by Paul Edwards.

2.5. The DCC attempts to account for this inconstancy by explaining that it is due to multiple injections of a single aliquot. In other words, the steroids extracted from one 4 ml sample of urine were analysed multiple times. This, however, was not the procedure described by Professor Cowan in his letter of 2nd July 1997. Also, he has openly admitted in more recent correspondence that he had:

“…not checked the number of measurements conducted in this case…” (9th December 2004).

2.6. The tension between Professor Cowan’s assertion that some aliquots were analysed a number of times and his assertion that he had not checked the number of measurements, can be resolved by inspecting the relevant intra-laboratory chain of custody and aliquot forms. The Edwards’ team has asked for sight of these since the initial hearing; our request has consistently been refused. The data which we wish to see will show precisely how much urine was used for each procedure carried out in the laboratory.

2.7. It should be noted that the IOC require this data to be recorded:

“Aliquots and intra laboratory chain of custody forms shall be used by laboratory personnel for conducting the initial and confirmatory tests” [Appendix d, sub-section 1.1, point B: Olympic movement Anti-Doping Code].

2.8. If this data is not released, then we can only rely on the evidence at hand: namely, Professor Cowan’s written description of the method which was used (his letter of 2nd July 1997) and the results from fifty-one-plus analyses.

3. Review of A sample analyses of sample 010830

Number of analyses: 10

Dates of analysis: 25th June, 1 sample; 26th June, 2 samples; 28th June, 7 samples.

Number of standards run: 0

3.1. Sample A010830 was first analysed on 25th June 1997. The A sample container was discarded after opening and cannot be used as evidence. It should be observed in passing that discarding the sample containers was contrary to IOC best practice:

“Specimen bottles and original chain of custody forms will normally be retained within the accession area until all analyses have been completed” [Appendix d, subsection 1.1, point B: Olympic movement ant-doping code].

3.2. A second analysis of the A sample was performed on 26th June. Also, and without obvious reason, a further two analyses of the A sample were conducted on 17th July: one at 10:56am; the second at 1:57 pm. These samples (tab 34, 35 36 and 37 of the bundle) are of particular interest because they have been printed twice, with peak areas and heights differing between replicate print-outs. These variations cannot have been created by transcription errors because they are direct print-outs from the Mass Spectrometer control software. The variations can therefore only be ascribed to an operator reprocessing the sample data file. It is quite possible that there was a valid reason for the reprocessing (for example, the automated peak detection software may not have functioned correctly); however, no record of any reprocessing or methods and justification for reprocessing exist. The reprocessing could therefore have resulted in a negative being turned into a positive. Accordingly, we cannot be certain about the validity of the results. The DCC intra-laboratory chain of custody forms - which should have recorded and accounted for the reprocessing - should be released. Unfortunately the DCC has declined to do this. The integrity of the results has therefore not been established.

3.3. Subsequent to the above analyses, a further seven A samples where analysed on 28th June. Although replicate analysis is always good, it is puzzling why ten replicate analyses of the A sample were performed over a three day period. No notes, explanation or intra-laboratory chain of custody have been provided by the DCC to explain this highly unusual practice.

4. Poor Identification of T and ET in the A sample

4.1. No external standards appear to have been run during any of the A sample analyses which I have listed above. It is therefore impossible to identify which of the numerous peaks within the Mass Chromatograms represent T and ET.

4.2. Further to this point, the Mass Spectrum for the individual peaks thought to be T and ET do not exist, again making identification of the correct compounds impossible.

4.3. The booking form (tab 1 page 47) does refer to a GC-MS-MS confirmation of T, but notes difficulties in identifying ET due to the low concentration. Accordingly, it is distinctly possible that the ET peak has been misidentified. With no standards or mass spectral data for the individual peak, it is impossible to determine what the true T/ET ratio of the sample is.

4.4. Professor Cowen has stated that the addition of 3dT and 3dET to each sample is sufficient to enable identification of the related T and ET peaks. This argument is unconvincing because without mass spectral data it is impossible to determine which of the numerous peaks are 3dT / 3dET, let alone which were Paul Edwards’ endogenous T and ET. Further data, which I will discuss later, will show that large discrepancies have occurred in retention time, so even this rudimentary identification method appears to be invalid (see Appendix A & B for a brief description on retention times and compound identification).

5. Review of further analysis of urine 010830 (A and B)

Number of A sample analyses: 41

Dates of analysis: 18th to 20th September

Number of standards run: 6

5.1. The B sample was first analysed on the 18th September. It should be noted that the B sample container was damaged and had to be opened with a hacksaw, the A sample container having already been discarded by the laboratory (the validity of which I have discussed in section 3).

6. Poor Identification of T and ET in samples run between the 18th and 20th September

6.1. As was the case in the A sample analyses, complete mass spectral data has been withheld for the peaks suspected of being T and ET. As a result, the T and ET peaks have to be identified during every sample analysis solely on the basis of retention time (ten or more peaks are present in every sample even when monitoring the 432 ion). This approach is highly unsatisfactory and would not be acceptable in other laboratories.

6.2. Pages 51 to 62 (tabs 2 and 3) display GC-MS-MS analysis, including full mass spectrums for T and ET. I assume this analysis has been performed on the samples in question for the purpose of identifying T and ET. It is notable that the T and ET retention time of a “20 to 1” standard run on the GC-MS-MS system differs significantly from the retention times observed in the analysis of urine samples on the standard mass spectrometer (GC-MS). During GC-MS-MS analysis, whilst monitoring the 432 ion, the retention time of T was approximately 9 minutes 50 seconds. When run on a GC-MS, the average retention time was 12 minutes 40 seconds (sd 0.04 n = 41). It is also notable that the ET peak could not be identified even by the far superior and more sensitive analytical method of GC-MS-MS. This means that we can have no confidence in the identification of either the ET or T peaks on any of the subsequent analysis. All the analyses run on sample 010830 must therefore be considered invalid.

(See Appendix A & B for a brief description on retention times and compound identification.)

7. Lack of calibration data for all sample analysis

7.1. When we arrived at the DCC we were informed that full calibrations curves had been run but, due to their commercial sensitivity, we were not permitted to view this data. It should be pointed out that a clerical member of staff and not a competent mass spectrometrist provided this information. I also point out that in more recent T/ET ratio cases, the DCC has freely provided the standard calibration curves. An example is provided in Appendix C. This inconsistency is unacceptable.

7.2. Despite this, even if the calibration curves do exist, serious questions about their validity remain unanswered.

7.3. Each print-out of the sample analysis has a time-stamp and instrument identification. From this information it is possible to determine the time line of when and on which instrument the samples where analysed. From this data it is clear that the B samples where run in two batches. The first batch commenced at 2:58am on 18th September and ran for 8 hours and 24 minutes. The second batch commenced at 2:08pm on 19th September and ran for 17 hours. During the combined time of 25 hours we can be certain that no calibration curves were run on the instrument in question. This can be confirmed because each urine analysis was carried out concurrently, with insufficient time between analyses for any standards to be run.

7.4. Even if the calibration curve had been run immediately prior to, or following, the analyses of the samples, the time elapsed would make these curves at best inaccurate and at worst invalid.

7.5. The need to run standards within an analytical batch is well-illustrated by looking at the peak areas of the deuterated internal standards of each analysis (Figure 1a & b). The purpose of the internal standard is to correct variations in the efficiency of steroid extraction from the urine between different samples (see Appendix B for further details). Small variations in the concentration of the internal standard are only to be expected (20% is not unusual). In Paul Edwards’ case, however, the variation is extremely large, with d3T varying by up to 300% and d3ET varying by up to 200% (Figure 1). To put this in perspective, it means that the amount of steroid extracted from the urine (for subsequent analysis) could have varied from 30% to 90% efficiency. This is self-evidently unacceptable.

(See Appendix B for a brief description on compound identification and quantification and the use of internal and external standards.)

Figure 1a. Changes in Tri Deuterated Testosterone peak areas during analysis of urine samples between the 19th and 20th of September. All data has been ranked from smallest to largest areas (7 outliers have been removed).

Figure 1b. Changes in Tri Deuterated Epitestosterone peak areas during analysis of urine samples between the 19th and 20th of September. All data has been ranked from smallest to largest areas (6 outliers have been removed).

8. Improper use of 20 to 1 standard

8.1. Within the two batches of samples run between 18th and 20th September, a single point standard has been run six times. This standard is labelled T:ET STD 20:1 (I assume this standard contains T and ET with a ratio of 20 to 1). This standard does not appear to be used in the calculation of the urinary concentrations and may simply represent a quality control measure. However, there are two worrying features displayed by this standard, both of which draw into question the validity of the urine analysis.

Loss of the deuterated component.

8.2. As with urine samples, standards also require the presence of a second internal deuterated component. This ensures that any changes in extraction efficiency between standards, and samples, are corrected. However, in the six “20 to 1” standards which were run, the deuterated internal standard is missing. The DCC operator comments on the absence of this deuterated peak on the print out of analysis 16. The comment, in pen, simply states “No D3 M1” and is dated the 22nd of September.

8.3. The absence of the deuterated peak in the only standard run within the batch would normally result in all other analyses in the batch being discarded. I would be interested to know what the laboratory’s standard operating procedures state about this eventuality. I also believe that the UKAS and GLP accrediting authorities would be interested in the fact that one of their members was releasing results in such circumstances.

8.4. In recent correspondence, Professor Cowan has stated that:

“The absence of the deuterated internal standards in the ’20:1’ standards was at the request of Mr Edwards’ scientific expert, Dr John Honour. It is my understanding that Dr Honour was satisfied with these data”.

8.5. Whether or not the independent witness was satisfied with the lack of an internal deuterated standard is irrelevant (and I suggest that Dr Honour is approached to confirm this statement). The most important matter is whether this procedure is right or wrong. Professor Cowan’s comment is deficient in not addressing this. The absence of an internal deuterated component in the only standard provided is most certainly wrong. Moreover, it should be noted that Professor Cowan does not state that the absence of the deuterated standard was correct.

The presence of more than two peaks.

8.6. The other concerning aspect of the six standards is the appearance of a third peak in a two compound standard (there should be one peak corresponding to each compound in the standard). The second peak appears on the Mass Chromatogram around the same time as ET and with the same mass charge ratio (the chemical fingerprint). As such it is not possible to distinguish between the standard and the “ghost peak”.

8.7. There are three possible origins of the “ghost peak”.

It could be due to carry-over from the previous analysis (memory effect).

It could be due to contamination.

It could be due to poor derivitization of the standard.

8.8. Memory effect. It is highly unlikely that the unknown peak has come from carry-over or memory effect. This is because no peaks were present with similar retention times in previous analysis. This can be seen in Table 1.

Table 1. Comparison between the retention times of the unknown peak in standard “T:ET STD 20:1”, compared to retention times from peaks within the sample analysed immediately prior to the standard

Standard “Run” number

Elution Time of unknown peak

Elution time of closest peak within the previous sample prior to the elution time of the unknown peak

Elution time of closest peak within the previous sample after the elution time of the unknown peak

16

12.471

12.456

12.545

24

12.471

12.355

12.546

37

12.465

12.455

12.539

8.9. Poor Derivitization. If the compound is poorly derivitized it is common to see extra “ghost” peaks appearing in the Chromatogram (see Appendix A for a brief description of derivitisation). This can be due to the creation of isomers of the original compound. It is perfectly reasonable to assume this may have occurred in this instance. However, it should be noted that ET is an isomer of T; thus, if poor derivitization resulted in isomerisation it is likely that both the T and the ET would undergo the same process. It might, therefore, be argued that four peaks should be present in the Mass Chromatogram. However, due to the greater abundance of T than ET, it is quite possible that only isomerised T would be visible on the scan. I believe the unknown peak may be isomerised T due to its elution at a similar time of the known T isomer, ET. The slight difference in retention time probably reflects a slightly different form of the T isomer.

8.9. The implications of this are significant because the preparation and derivitization process for the urine samples and the standards should have been identical. In this case, it is almost certain that poor derivitization of the samples will also have occurred. The identification of such errors is one of the reasons standards are run.

8.10. If within a batch run the standards are not of high quality, the validity of all analyses performed under the same conditions is highly questionable and should certainly not be used to destroy an athletes’ career.

8.11. When questioned on this matter, Professor Cowen chose not to comment.

8.12. Contamination. Although I believe the real cause of the unknown peak is poor derivitization, it is still possible that the “ghost peak” could be present due to contamination. If this is the case, then it is quite possible that the urine samples, which were processed in the same manner and with the same apparatus as the standards, also contain contaminants. If this is true, it follows that the analysis is flawed and should be disregarded.

9. Large Tri-deuterated peaks in Water Blank analysis

9.1. An injection of a pure compound, such as water or a similar solvent, is normally run in a batch analysis to ensure that only minimal contamination occurs from carry-over or memory from a previous sample analyses. In other words, if the analysis was perfect, a pure water sample would contain no peaks whatsoever when analysed in the same manner as the urine samples and the standards.

9.2. In the case at hand, we see some peaks in the water blanks when monitoring ion 432, some of which correspond to T and ET. It is therefore evident that there is some carry-over from the previous analysis. This is not necessarily a problem because you are always likely to get some contamination from the previous sample. I believe that, in the case in hand, the level of contamination whilst monitoring the 432 ion will not have significantly changed the analytical results.

9.3. This, however, cannot be said when monitoring the 435 ion (this monitors the deuterated component of the sample i.e. the internal standard). Within every water blank, at the same retention times as Deuterated T (d3T) and Deuterated ET (d3Et), large deuterated peaks are observed. These peaks are over 100% larger in area and height than observed in samples known to have been spiked with a deuterated standard (Table 2). As the peaks are over twice the height of the samples run prior to the blank, they cannot be due to carry over.

9.4. The presence of these peaks are hard to explain but because the deuterated standards will have been used to calculate the concentrations of the T and ET in the urine samples the huge peaks in a “water blank” must invalidate the entire analysis as presented. Even if deuterated standards were placed in the water blanks deliberately (which would beg the question why they were labelled “water blank”), this would have prevented any measure of contamination from previous samples. Again this would invalidate the entire analysis as presented.

9.5. In recent correspondence, Professor Cowan argues that the deuterated standards were added to the water in order to determine whether the d3T and d3TE were somehow converting to T and ET (something which would be highly unlikely). If this were the case, I ask the simple question as to why the sample was labelled “Water blank” and not “Water blank plus tri-deuterated standards” (or appropriately abbreviation). I also ask why such a procedure has not been reported in any of the descriptions or methods provided to date. This raises a simple question: were any other compounds added to the samples or standards which do not appear on their labels or described within their methods?

Conclusion

C.1. The analysis of urine sample 010830 raises many questions. These range from why the sample had to be opened with a hacksaw to whether sufficient urine was available to carry out the analysis presented. There is also a very clear error in the calculations of T/ET ratio, with the calculation sheet clearly placing the 432 ion results into the 435 ion column.

C.2. Combined with this, the method of calibration is clearly not standard and has been withheld from us, even though it has recently been made available readily to other athletes. The few standards which have been run are clearly missing a deuterated component, thus making them useless. Compounding these problems, the standards also show clear evidence of contamination and/or poor derivitization.

C.3. Contamination also appears in the “Water Blanks”, with a sample of pure water (which should contain nothing other than H2O) presenting deuterated T and ET in concentration twice the amount present in samples deliberately spiked with these components.

C.4. In short, if the laboratory is unable to provide valid explanations and documented proof of the cause of these major anomalies, then all results from the sample should be withdrawn.

Appendix A

A brief introduction to Gas Chromatography Mass Spectrometry

AA1. The analytical technique typically used to identify and quantify steroids within an athlete’s urine is known as Gas Chromatography Mass Spectrometry (GC-MS). This was the technique used by the DCC to identify T and ET in Paul Edwards’ case.

AA2. This form of analysis requires the analyte to be volatile; i.e. capable of converting from a solid to a gas at low temperatures. Steroids are, however, semi or non-volatile substances, making GC-MS analysis of the original molecule difficult, if not impossible. It is therefore necessary to alter the chemical structure of the steroids to produce a secondary, more volatile, compound. This process is known as derivitization.

AA3. Once derivatized, the steroids within the sample are separated into individual compounds by Gas Chromatography (GC). GC separation requires vaporizing the sample in a heated injection port. After the sample has been volatilized, the resultant gas is passed through a long cylindrical glass tube known as a column. The inside of the column is coated with a thin chemical film, capable of retarding the flow of specific compounds within the sample gas. The degree to which a compound’s flow is retarded depends on the physical and chemical properties of substance in question (the compound’s “polarity”).

AA4. As different compounds have different polarities, their retention time within the column also varies. This results in individual compounds eluting from the column at different time intervals (relative to the time the sample was injected into the column). The time at which individual compounds exit the column (elution time) can be plotted against their abundance, thus producing a chromatogram (Figure 1).

Figure 1. An example of a chromatogram from a sample containing four compounds all of which have been resolved into four separate peaks following gas chromatography (GC).

AA5. Chromatograms are useful diagnostic tools; however, many compounds share similar retention times and retention times may also change. As a result chromatographic analysis alone is insufficient for conclusive identification of unknown substances. The application of chromatographic separation is instead primarily used to isolate individual compounds thus facilitating further Mass Spectrometry (MS) analysis.

AA5. As the separated compounds elute from the GC they are transported via a fused silica tube to the MS. Within the MS the compounds flow into an Electron Impact (EI) source. As the analytes enter the source, they pass through a stream of electrons. The electron beam alters a portion of the analytes in two specific ways:

AA6. [1] The electron beam impacts the analytes and displaces one or more electrons from the outer shell of the compound. As electrons are negatively charged and the analyte is neutral, the removal of one or more electrons produces a positively charged particle, known as an ion.

AA7. [2] The second effect of the electron impact is to “crack” or breakdown a portion of the original parent analyte into smaller particles often referred to as daughter ions.

AA8. Once the parent and daughter ionic compounds are formed, the MS accelerates the ions through a mass filter to an amplifier and ultimately to a detector. The mass filter enables the mass (m) and charge (z) of the ions produced from the analyte to be determined.

AA9. The different m/zs of the ions and their relative abundance can then be used as a chemical fingerprint to enable identification of the original parent compound. This fingerprint is described graphically by means of a mass spectrum. A mass spectrum is simply a plot of the different ion’s m/z values against their abundance (as can be seen in Figure 2).

Figure 2. A simplified mass spectrum that would be obtained for a nandrolone metabolite. Under this protocol, a nandrolone metabolite would produce two major ions (405 & 420) and two diagnostic ions (225 & 315).

Appendix B

A brief introduction to the use of standards and instrument calibration

Ap.B.1. To validate the results given by the DCC, full records of the standards must be obtained. This includes both internal and external standards.

Internal standards

Ap.B.2. Internal standards are compounds which are added to an athlete’s urine before the steroids within the sample are extracted for subsequent GC-MS analysis. These standards are steroids (in this case T and ET) which can be distinguished from an athlete’s endogenous steroids by the addition of three heavy isotopes labels of hydrogen (otherwise known as deuterium). This produces “tri-deuterated” standards, i.e. tri-deuterated T and ET (otherwise written as 3dT and 3dET).

Ap.B.3.The purpose of these standards is to measure variations in the efficiency of steroid extraction between different samples (and different external standards). For example, we could have two urine samples (A and B), one of which is relatively clear (urine A) and one of which is relatively cloudy (urine B). The cloudy B sample may contain dead skin cells or other substances which will interfere with steroid extraction. A greater percentage of the steroids will therefore be extracted from the clearer urine A than the cloudy urine B. These differences can be significant with a 20% variation between urine samples not unusual. The internal standards are used to correct for these variations in a relatively simple process. Since you know how much internal standard you have added to the sample, you can compare this with how much you ultimately extract. This will give you a relatively precise measure of your extraction efficiency of the internal standard. It is then reasonable to assume that the extraction efficiency of the non-deuterated endogenous steroid is the same as that for the internal standard. Simple arithmetic can then be used to correct any variations in extraction efficiency.

External Standards

Ap.B.4. External standards are standards that are run separately, or external, from the urine samples in question. These standards are used to enable quantification and identification of the compounds in the urine.

Ap.B.5. Identification of a compound is critical in any doping case. Although the procedure is relatively simple, its implementation is often poorly handled. In simple terms, if you wish to identify T and ET within a urine sample you must run a T and ET standard contemporaneously and on the same MS as the urine. The urinary T / ET and standard T / ET must share two traits in common for positive identification:

Similar GC retention times.

Identical Mass Spectrums.

(If you are not clear what these terms mean, please refer to Appendix A.)

Ap.B.6. As stated by the IOC medical code:

“…all the information collected …must be… in agreement with the known facts and structure of the doping agent or metabolite(s)”. (IOC Medical Code, Appendix D, Section 2.1.4.).

Ap.B.7. You will have noted in this report that at best only one of these indicators appears to have been used by the DCC in Mr. Edwards’ case.

Ap.B.8. Quantification is a more complex procedure and can be achieved in a number of different ways. Unfortunately, the DCC has refused to reveal the way in which they quantified T and ET in Mr. Edward’s case, quoting commercial confidentiality as the reason. It should be noted that the DCC has provided this information freely to other T cases! However, the standard method of quantification, and method used by the DCC in other T cases involves running a series of six or more standards containing a range of concentrations. To demonstrate this is a standard procedure it is useful to refer to EEC regulations on steroid analysis that state:

“A minimum of six calibration points is required, adequately distributed along the calibration curve.” Section 2.3.1.2, 93/256/EEC.

Ap.B.9. When these samples are passed through a mass spectrometer they will produce a range of response from the MS detector in amps. These can then be plotted against the known concentrations of the standards to give a calibration curve (see Figure 1, next page).

ApB.10. Once this calibration curve is produced, subsequent urine samples (the T and ET concentrations of which are unknown) can be quantified using the regression analysis shown in Figure 1. In other words, the MS will give you a value in amps. This should be read off the Y-axis of the graph and a line extended from this point until it intercepts the calibration curve. The value on the X-axis at this point gives the concentration of T in ng/ml.

ApB.11. Despite our best efforts over the past six years, this very basic analytical data has not been provided.

Figure 1. Idealized calibration curve run a GC-MS system which is perfectly linear. The curve has a perfect regression fit (r2 = 1) a slope of 1 and a Y intercept of 0.

Appendix C"

Report wrote:

Lean Cuisine wrote:

"Review of new analytical data withheld from Paul Edwards and the BAF/UKA disciplinary and appeals panels

Just a reminder as has not been posted for a while.

Simon Davis PhD

INTRODUCTION

Note to reader

To understand fully the issues relating to the analytical data which was withheld from Paul Edwards’ team, it is first necessary to understand the basic principles of the analytical technique used by the Drugs Control Centre to analyze Paul Edwards urine sample. The technique in question is known as Gas Chromatography Mass Spectrometry (GC-MS). It is a well-recognized method of identifying and quantifying steroids or steroid breakdown products (metabolites) in an athlete’s urine. If the reader of this report is not familiar with this technique, I suggest that he/she reads the brief introduction held in Appendices A and B.

Int.1. On the 09 of May 2002, Mr. Caborn, Minister for Sport, stated in the House of Commons in the Adjournment Debate which was introduced by Andrew Hunter MP that Paul Edwards had received all the data relating to his doping case. It has subsequently been discovered that this was not the case: a further 600 pages of undisclosed analytical data had been withheld by the Drugs Control Centre (DCC). Of course, the Minister did not knowingly mislead the House; however, the briefing which he received prior to the Adjournment Debate from UK Sport, his officials and perhaps other advisers clearly did not include the data which had been withheld by the DCC.

Int.2. This data had not only been withheld from the Minister and Paul Edwards’ team but also from each of the disciplinary and appeals panels before which the case had been brought. A life-ban was imposed on Paul Edwards without any awareness of the data which is discussed in this report.

Int.3. Paul Edward’s team has been aware of the existence of this data since the outset of the case against him. We have campaigned vigorously for its full disclosure. Obstacles have been put in the way of our obtaining. We have met with prevarication and delaying tactics.

Int.4. We requested in writing prior before the first hearing that we could have sight of all data. Our request was unsuccessful. Subsequently, we have submitted several requests for data under the Data Protection Act to the British Athletics Federation (BAF), UK Sport, UK Athletics (UKA) and the DCC. Despite its statutory obligation, the DCC refused to disclose any data to us. We therefore made a formal complaint to the Data Protection Agency, which sent a formal written warning to the DCC. The DCC ignored this warning and continued to withhold the relevant data. The DCC claimed that because the data was held on ‘continuous feed paper’ (i.e. all the sheets were joined together), it would require “disproportionate effort” to photocopy.

Int.5. In order to gain access to the data, two legal representatives and I used the Data Protection Act to gain access to the DCC laboratory in order to view the data which “required a disproportionate effort” to copy. Despite the paper being joined together, it took one person around thirty minutes to photocopy the entire bundle (significantly less time than it took to collect and record Paul Edward’s urine sample!). The effort invested by the DCC to prevent the release of this data over the previous five years had been significantly greater than the thirty minutes of “disproportionate effort” which would have been required to photocopy the bundle.

Int.6. On initially looking through the bundle, we noticed that it included a number of hand-written directions. A blue post-it tab on the top of the bundle was particularly interesting. It read, “Just the T/E sheet; don’t give attached data”. At this stage it is relevant to recall the International Olympic Committee (IOC) ruling:

“Copies of all analytical results shall be made available from the laboratory when requested by an appropriate authority” [Appendix d, sub-section 1.2 Olympic Movement Anti-Doping Code].

It appears that the DCC feels that it is not obliged to follow its own regulations or inform Ministers of State to the existence of data, if it does not wish to.

Int.7. A review of the data in the bundle quickly established that yet further data is still being withheld by the DCC. Crucially, this included the results from the standards and calibrations used to identify and quantify any Testosterone (T) and Epitestosterone (ET) in Paul Edward’s sample. Withholding this data makes it impossible to validate the DCC’s findings or identify any possible errors in the DCC’s results (as occurred in the DCC versus Mark Hylton). In addition, the results of Paul Edwards Luteinizing Hormone (LH) tests are also still being withheld, despite being used as part of the evidence of Mr. Edward’s guilt. Furthermore, the results summary form (tab 6) refers to a further two GC-MS analyses for which the data is again absent, whilst tab 1 of the bundle refers to a further GC-MS-MS analysis whose data is also missing!

Int.8. It is clear that crucial data which indisputably relates to the Edwards’ case is, to this day, being withheld by the DCC in contravention of the Data Protection Act. If the case against Paul Edwards is sound, one can conceive no reason why this data should not have been released immediately in order to validate the DCC’s finding

Review of new data evidence obtained from the DCC

Int.9. All points and comments below relate to the new evidence which was withheld from Paul Edwards’ team and from the relevant disciplinary and appeals panels.

Int.10. After the reviewing the new data from the DCC, a number of serious questions were raised over the validity of the analysis which had been made of Edwards’ sample [010830]. I address these in this paper in the order listed below:

Apparent use of incorrect ion to calculate Paul Edwards’ T/ET ratio.

Inconstancies in the volumes of urine used in the analysis of sample 010830.

Review of A sample analysis of urine 010830.

Poor identification of T and ET in the A sample.

Review of further analysis of urine 010830 (A and B).

Poor identification of T and ET in samples run between the 18th and 20th of September.

Lack of calibration data for all sample analysis.

Improper use of 20 to 1 standard.

Large tri-deuterated peak in Water Blank analysis.

1. Apparent use of incorrect ion to calculate Paul Edwards’ T/ET ratio

1.1. Within tab 6 of the bundle which we obtained from the DCC, a results summary/calculation table is presented. Although the DCC have refused to explain the method of calibration, there is clearly a gross error present within this sheet. The values used for the peak heights of T are labelled as having been measured using the ion 435 (m/z 435.3). This ion is in fact the tri-deuterated internal T standard (3dT) and not the T from Paul Edwards’ sample.

1.2. If the T of the athlete’s urine has been mistakenly confused with the internal standard, then clearly the reported T/ET ratio becomes meaningless. This rather basic error draws into question the validity of the whole test and also invites the question: what other errors have been made?

1.3. Professor Cowan has commented on this point. He describes it as a minor “transcription” error. However, despite four Data Protection searches, we have found no documentary evidence that this error had been either identified or rectified prior to the T/E ratios being released to the disciplinary panels. Two conclusions can be drawn: either that an error went uncorrected and all the reported T/E ratios are wrong; or that the error was corrected and, despite four Data Protection requests, documentary evidence which establishes that corrections were made have been withheld.

1.4. Paul Edward’s team has been unable to investigate this point further as the calibration methodology and data necessary to do so has been withheld by the DCC. If this is a minor transcription error, we are bewildered as to why the DCC refuses to release the data which will prove it so.

2. Inconsistencies in the volumes of urine used in analysing sample 010830

2.1. Professor Cowan describes the method for sample preparation in his letter of 2nd July 1997. He stated that each extraction required 4 ml of urine.

2.2. Within the bundle supplied by the DCC, results from fifty-one GC-MS analyses of sample 010830 were printed. The bundle also contained print-outs of a further three HPLC analysis (tab 29) and at least one GC-MS-MS analysis (tab 3). A further GC-MS-MS analysis is referred to in the event log (tab 1) but the relevant data is missing. Also, the results summary form (tab 6) refers to a further two GC-MS analyses for which the data is also missing.

2.3. If the DCC had followed its own written protocol and used 4 ml of urine for each extraction, over 200 ml of urine would have been required to perform fifty-one GC-MS analyses. Bear in mind that this excludes the HPLC, GC-MS-MS and missing GC-MS-MS and GC-MS analysis.

2.4. From the doping control booking form it is clear that only 170 ml of urine were provided by Paul Edwards.

2.5. The DCC attempts to account for this inconstancy by explaining that it is due to multiple injections of a single aliquot. In other words, the steroids extracted from one 4 ml sample of urine were analysed multiple times. This, however, was not the procedure described by Professor Cowan in his letter of 2nd July 1997. Also, he has openly admitted in more recent correspondence that he had:

“…not checked the number of measurements conducted in this case…” (9th December 2004).

2.6. The tension between Professor Cowan’s assertion that some aliquots were analysed a number of times and his assertion that he had not checked the number of measurements, can be resolved by inspecting the relevant intra-laboratory chain of custody and aliquot forms. The Edwards’ team has asked for sight of these since the initial hearing; our request has consistently been refused. The data which we wish to see will show precisely how much urine was used for each procedure carried out in the laboratory.

2.7. It should be noted that the IOC require this data to be recorded:

“Aliquots and intra laboratory chain of custody forms shall be used by laboratory personnel for conducting the initial and confirmatory tests” [Appendix d, sub-section 1.1, point B: Olympic movement Anti-Doping Code].

2.8. If this data is not released, then we can only rely on the evidence at hand: namely, Professor Cowan’s written description of the method which was used (his letter of 2nd July 1997) and the results from fifty-one-plus analyses.

3. Review of A sample analyses of sample 010830

Number of analyses: 10

Dates of analysis: 25th June, 1 sample; 26th June, 2 samples; 28th June, 7 samples.

Number of standards run: 0

3.1. Sample A010830 was first analysed on 25th June 1997. The A sample container was discarded after opening and cannot be used as evidence. It should be observed in passing that discarding the sample containers was contrary to IOC best practice:

“Specimen bottles and original chain of custody forms will normally be retained within the accession area until all analyses have been completed” [Appendix d, subsection 1.1, point B: Olympic movement ant-doping code].

3.2. A second analysis of the A sample was performed on 26th June. Also, and without obvious reason, a further two analyses of the A sample were conducted on 17th July: one at 10:56am; the second at 1:57 pm. These samples (tab 34, 35 36 and 37 of the bundle) are of particular interest because they have been printed twice, with peak areas and heights differing between replicate print-outs. These variations cannot have been created by transcription errors because they are direct print-outs from the Mass Spectrometer control software. The variations can therefore only be ascribed to an operator reprocessing the sample data file. It is quite possible that there was a valid reason for the reprocessing (for example, the automated peak detection software may not have functioned correctly); however, no record of any reprocessing or methods and justification for reprocessing exist. The reprocessing could therefore have resulted in a negative being turned into a positive. Accordingly, we cannot be certain about the validity of the results. The DCC intra-laboratory chain of custody forms - which should have recorded and accounted for the reprocessing - should be released. Unfortunately the DCC has declined to do this. The integrity of the results has therefore not been established.

3.3. Subsequent to the above analyses, a further seven A samples where analysed on 28th June. Although replicate analysis is always good, it is puzzling why ten replicate analyses of the A sample were performed over a three day period. No notes, explanation or intra-laboratory chain of custody have been provided by the DCC to explain this highly unusual practice.

4. Poor Identification of T and ET in the A sample

4.1. No external standards appear to have been run during any of the A sample analyses which I have listed above. It is therefore impossible to identify which of the numerous peaks within the Mass Chromatograms represent T and ET.

4.2. Further to this point, the Mass Spectrum for the individual peaks thought to be T and ET do not exist, again making identification of the correct compounds impossible.

4.3. The booking form (tab 1 page 47) does refer to a GC-MS-MS confirmation of T, but notes difficulties in identifying ET due to the low concentration. Accordingly, it is distinctly possible that the ET peak has been misidentified. With no standards or mass spectral data for the individual peak, it is impossible to determine what the true T/ET ratio of the sample is.

4.4. Professor Cowen has stated that the addition of 3dT and 3dET to each sample is sufficient to enable identification of the related T and ET peaks. This argument is unconvincing because without mass spectral data it is impossible to determine which of the numerous peaks are 3dT / 3dET, let alone which were Paul Edwards’ endogenous T and ET. Further data, which I will discuss later, will show that large discrepancies have occurred in retention time, so even this rudimentary identification method appears to be invalid (see Appendix A & B for a brief description on retention times and compound identification).

5. Review of further analysis of urine 010830 (A and B)

Number of A sample analyses: 41

Dates of analysis: 18th to 20th September

Number of standards run: 6

5.1. The B sample was first analysed on the 18th September. It should be noted that the B sample container was damaged and had to be opened with a hacksaw, the A sample container having already been discarded by the laboratory (the validity of which I have discussed in section 3).

6. Poor Identification of T and ET in samples run between the 18th and 20th September

6.1. As was the case in the A sample analyses, complete mass spectral data has been withheld for the peaks suspected of being T and ET. As a result, the T and ET peaks have to be identified during every sample analysis solely on the basis of retention time (ten or more peaks are present in every sample even when monitoring the 432 ion). This approach is highly unsatisfactory and would not be acceptable in other laboratories.

6.2. Pages 51 to 62 (tabs 2 and 3) display GC-MS-MS analysis, including full mass spectrums for T and ET. I assume this analysis has been performed on the samples in question for the purpose of identifying T and ET. It is notable that the T and ET retention time of a “20 to 1” standard run on the GC-MS-MS system differs significantly from the retention times observed in the analysis of urine samples on the standard mass spectrometer (GC-MS). During GC-MS-MS analysis, whilst monitoring the 432 ion, the retention time of T was approximately 9 minutes 50 seconds. When run on a GC-MS, the average retention time was 12 minutes 40 seconds (sd 0.04 n = 41). It is also notable that the ET peak could not be identified even by the far superior and more sensitive analytical method of GC-MS-MS. This means that we can have no confidence in the identification of either the ET or T peaks on any of the subsequent analysis. All the analyses run on sample 010830 must therefore be considered invalid.

(See Appendix A & B for a brief description on retention times and compound identification.)

7. Lack of calibration data for all sample analysis

7.1. When we arrived at the DCC we were informed that full calibrations curves had been run but, due to their commercial sensitivity, we were not permitted to view this data. It should be pointed out that a clerical member of staff and not a competent mass spectrometrist provided this information. I also point out that in more recent T/ET ratio cases, the DCC has freely provided the standard calibration curves. An example is provided in Appendix C. This inconsistency is unacceptable.

7.2. Despite this, even if the calibration curves do exist, serious questions about their validity remain unanswered.

7.3. Each print-out of the sample analysis has a time-stamp and instrument identification. From this information it is possible to determine the time line of when and on which instrument the samples where analysed. From this data it is clear that the B samples where run in two batches. The first batch commenced at 2:58am on 18th September and ran for 8 hours and 24 minutes. The second batch commenced at 2:08pm on 19th September and ran for 17 hours. During the combined time of 25 hours we can be certain that no calibration curves were run on the instrument in question. This can be confirmed because each urine analysis was carried out concurrently, with insufficient time between analyses for any standards to be run.

7.4. Even if the calibration curve had been run immediately prior to, or following, the analyses of the samples, the time elapsed would make these curves at best inaccurate and at worst invalid.

7.5. The need to run standards within an analytical batch is well-illustrated by looking at the peak areas of the deuterated internal standards of each analysis (Figure 1a & b). The purpose of the internal standard is to correct variations in the efficiency of steroid extraction from the urine between different samples (see Appendix B for further details). Small variations in the concentration of the internal standard are only to be expected (20% is not unusual). In Paul Edwards’ case, however, the variation is extremely large, with d3T varying by up to 300% and d3ET varying by up to 200% (Figure 1). To put this in perspective, it means that the amount of steroid extracted from the urine (for subsequent analysis) could have varied from 30% to 90% efficiency. This is self-evidently unacceptable.

(See Appendix B for a brief description on compound identification and quantification and the use of internal and external standards.)

Figure 1a. Changes in Tri Deuterated Testosterone peak areas during analysis of urine samples between the 19th and 20th of September. All data has been ranked from smallest to largest areas (7 outliers have been removed).

Figure 1b. Changes in Tri Deuterated Epitestosterone peak areas during analysis of urine samples between the 19th and 20th of September. All data has been ranked from smallest to largest areas (6 outliers have been removed).

8. Improper use of 20 to 1 standard

8.1. Within the two batches of samples run between 18th and 20th September, a single point standard has been run six times. This standard is labelled T:ET STD 20:1 (I assume this standard contains T and ET with a ratio of 20 to 1). This standard does not appear to be used in the calculation of the urinary concentrations and may simply represent a quality control measure. However, there are two worrying features displayed by this standard, both of which draw into question the validity of the urine analysis.

Loss of the deuterated component.

8.2. As with urine samples, standards also require the presence of a second internal deuterated component. This ensures that any changes in extraction efficiency between standards, and samples, are corrected. However, in the six “20 to 1” standards which were run, the deuterated internal standard is missing. The DCC operator comments on the absence of this deuterated peak on the print out of analysis 16. The comment, in pen, simply states “No D3 M1” and is dated the 22nd of September.

8.3. The absence of the deuterated peak in the only standard run within the batch would normally result in all other analyses in the batch being discarded. I would be interested to know what the laboratory’s standard operating procedures state about this eventuality. I also believe that the UKAS and GLP accrediting authorities would be interested in the fact that one of their members was releasing results in such circumstances.

8.4. In recent correspondence, Professor Cowan has stated that:

“The absence of the deuterated internal standards in the ’20:1’ standards was at the request of Mr Edwards’ scientific expert, Dr John Honour. It is my understanding that Dr Honour was satisfied with these data”.

8.5. Whether or not the independent witness was satisfied with the lack of an internal deuterated standard is irrelevant (and I suggest that Dr Honour is approached to confirm this statement). The most important matter is whether this procedure is right or wrong. Professor Cowan’s comment is deficient in not addressing this. The absence of an internal deuterated component in the only standard provided is most certainly wrong. Moreover, it should be noted that Professor Cowan does not state that the absence of the deuterated standard was correct.

The presence of more than two peaks.

8.6. The other concerning aspect of the six standards is the appearance of a third peak in a two compound standard (there should be one peak corresponding to each compound in the standard). The second peak appears on the Mass Chromatogram around the same time as ET and with the same mass charge ratio (the chemical fingerprint). As such it is not possible to distinguish between the standard and the “ghost peak”.

8.7. There are three possible origins of the “ghost peak”.

It could be due to carry-over from the previous analysis (memory effect).

It could be due to contamination.

It could be due to poor derivitization of the standard.

8.8. Memory effect. It is highly unlikely that the unknown peak has come from carry-over or memory effect. This is because no peaks were present with similar retention times in previous analysis. This can be seen in Table 1.

Table 1. Comparison between the retention times of the unknown peak in standard “T:ET STD 20:1”, compared to retention times from peaks within the sample analysed immediately prior to the standard

Standard “Run” number

Elution Time of unknown peak

Elution time of closest peak within the previous sample prior to the elution time of the unknown peak

Elution time of closest peak within the previous sample after the elution time of the unknown peak

16

12.471

12.456

12.545

24

12.471

12.355

12.546

37

12.465

12.455

12.539

8.9. Poor Derivitization. If the compound is poorly derivitized it is common to see extra “ghost” peaks appearing in the Chromatogram (see Appendix A for a brief description of derivitisation). This can be due to the creation of isomers of the original compound. It is perfectly reasonable to assume this may have occurred in this instance. However, it should be noted that ET is an isomer of T; thus, if poor derivitization resulted in isomerisation it is likely that both the T and the ET would undergo the same process. It might, therefore, be argued that four peaks should be present in the Mass Chromatogram. However, due to the greater abundance of T than ET, it is quite possible that only isomerised T would be visible on the scan. I believe the unknown peak may be isomerised T due to its elution at a similar time of the known T isomer, ET. The slight difference in retention time probably reflects a slightly different form of the T isomer.

8.9. The implications of this are significant because the preparation and derivitization process for the urine samples and the standards should have been identical. In this case, it is almost certain that poor derivitization of the samples will also have occurred. The identification of such errors is one of the reasons standards are run.

8.10. If within a batch run the standards are not of high quality, the validity of all analyses performed under the same conditions is highly questionable and should certainly not be used to destroy an athletes’ career.

8.11. When questioned on this matter, Professor Cowen chose not to comment.

8.12. Contamination. Although I believe the real cause of the unknown peak is poor derivitization, it is still possible that the “ghost peak” could be present due to contamination. If this is the case, then it is quite possible that the urine samples, which were processed in the same manner and with the same apparatus as the standards, also contain contaminants. If this is true, it follows that the analysis is flawed and should be disregarded.

9. Large Tri-deuterated peaks in Water Blank analysis

9.1. An injection of a pure compound, such as water or a similar solvent, is normally run in a batch analysis to ensure that only minimal contamination occurs from carry-over or memory from a previous sample analyses. In other words, if the analysis was perfect, a pure water sample would contain no peaks whatsoever when analysed in the same manner as the urine samples and the standards.

9.2. In the case at hand, we see some peaks in the water blanks when monitoring ion 432, some of which correspond to T and ET. It is therefore evident that there is some carry-over from the previous analysis. This is not necessarily a problem because you are always likely to get some contamination from the previous sample. I believe that, in the case in hand, the level of contamination whilst monitoring the 432 ion will not have significantly changed the analytical results.

9.3. This, however, cannot be said when monitoring the 435 ion (this monitors the deuterated component of the sample i.e. the internal standard). Within every water blank, at the same retention times as Deuterated T (d3T) and Deuterated ET (d3Et), large deuterated peaks are observed. These peaks are over 100% larger in area and height than observed in samples known to have been spiked with a deuterated standard (Table 2). As the peaks are over twice the height of the samples run prior to the blank, they cannot be due to carry over.

9.4. The presence of these peaks are hard to explain but because the deuterated standards will have been used to calculate the concentrations of the T and ET in the urine samples the huge peaks in a “water blank” must invalidate the entire analysis as presented. Even if deuterated standards were placed in the water blanks deliberately (which would beg the question why they were labelled “water blank”), this would have prevented any measure of contamination from previous samples. Again this would invalidate the entire analysis as presented.

9.5. In recent correspondence, Professor Cowan argues that the deuterated standards were added to the water in order to determine whether the d3T and d3TE were somehow converting to T and ET (something which would be highly unlikely). If this were the case, I ask the simple question as to why the sample was labelled “Water blank” and not “Water blank plus tri-deuterated standards” (or appropriately abbreviation). I also ask why such a procedure has not been reported in any of the descriptions or methods provided to date. This raises a simple question: were any other compounds added to the samples or standards which do not appear on their labels or described within their methods?

Conclusion

C.1. The analysis of urine sample 010830 raises many questions. These range from why the sample had to be opened with a hacksaw to whether sufficient urine was available to carry out the analysis presented. There is also a very clear error in the calculations of T/ET ratio, with the calculation sheet clearly placing the 432 ion results into the 435 ion column.

C.2. Combined with this, the method of calibration is clearly not standard and has been withheld from us, even though it has recently been made available readily to other athletes. The few standards which have been run are clearly missing a deuterated component, thus making them useless. Compounding these problems, the standards also show clear evidence of contamination and/or poor derivitization.

C.3. Contamination also appears in the “Water Blanks”, with a sample of pure water (which should contain nothing other than H2O) presenting deuterated T and ET in concentration twice the amount present in samples deliberately spiked with these components.

C.4. In short, if the laboratory is unable to provide valid explanations and documented proof of the cause of these major anomalies, then all results from the sample should be withdrawn.

Appendix A

A brief introduction to Gas Chromatography Mass Spectrometry

AA1. The analytical technique typically used to identify and quantify steroids within an athlete’s urine is known as Gas Chromatography Mass Spectrometry (GC-MS). This was the technique used by the DCC to identify T and ET in Paul Edwards’ case.

AA2. This form of analysis requires the analyte to be volatile; i.e. capable of converting from a solid to a gas at low temperatures. Steroids are, however, semi or non-volatile substances, making GC-MS analysis of the original molecule difficult, if not impossible. It is therefore necessary to alter the chemical structure of the steroids to produce a secondary, more volatile, compound. This process is known as derivitization.

AA3. Once derivatized, the steroids within the sample are separated into individual compounds by Gas Chromatography (GC). GC separation requires vaporizing the sample in a heated injection port. After the sample has been volatilized, the resultant gas is passed through a long cylindrical glass tube known as a column. The inside of the column is coated with a thin chemical film, capable of retarding the flow of specific compounds within the sample gas. The degree to which a compound’s flow is retarded depends on the physical and chemical properties of substance in question (the compound’s “polarity”).

AA4. As different compounds have different polarities, their retention time within the column also varies. This results in individual compounds eluting from the column at different time intervals (relative to the time the sample was injected into the column). The time at which individual compounds exit the column (elution time) can be plotted against their abundance, thus producing a chromatogram (Figure 1).

Figure 1. An example of a chromatogram from a sample containing four compounds all of which have been resolved into four separate peaks following gas chromatography (GC).

AA5. Chromatograms are useful diagnostic tools; however, many compounds share similar retention times and retention times may also change. As a result chromatographic analysis alone is insufficient for conclusive identification of unknown substances. The application of chromatographic separation is instead primarily used to isolate individual compounds thus facilitating further Mass Spectrometry (MS) analysis.

AA5. As the separated compounds elute from the GC they are transported via a fused silica tube to the MS. Within the MS the compounds flow into an Electron Impact (EI) source. As the analytes enter the source, they pass through a stream of electrons. The electron beam alters a portion of the analytes in two specific ways:

AA6. [1] The electron beam impacts the analytes and displaces one or more electrons from the outer shell of the compound. As electrons are negatively charged and the analyte is neutral, the removal of one or more electrons produces a positively charged particle, known as an ion.

AA7. [2] The second effect of the electron impact is to “crack” or breakdown a portion of the original parent analyte into smaller particles often referred to as daughter ions.

AA8. Once the parent and daughter ionic compounds are formed, the MS accelerates the ions through a mass filter to an amplifier and ultimately to a detector. The mass filter enables the mass (m) and charge (z) of the ions produced from the analyte to be determined.

AA9. The different m/zs of the ions and their relative abundance can then be used as a chemical fingerprint to enable identification of the original parent compound. This fingerprint is described graphically by means of a mass spectrum. A mass spectrum is simply a plot of the different ion’s m/z values against their abundance (as can be seen in Figure 2).

Figure 2. A simplified mass spectrum that would be obtained for a nandrolone metabolite. Under this protocol, a nandrolone metabolite would produce two major ions (405 & 420) and two diagnostic ions (225 & 315).

Appendix B

A brief introduction to the use of standards and instrument calibration

Ap.B.1. To validate the results given by the DCC, full records of the standards must be obtained. This includes both internal and external standards.

Internal standards

Ap.B.2. Internal standards are compounds which are added to an athlete’s urine before the steroids within the sample are extracted for subsequent GC-MS analysis. These standards are steroids (in this case T and ET) which can be distinguished from an athlete’s endogenous steroids by the addition of three heavy isotopes labels of hydrogen (otherwise known as deuterium). This produces “tri-deuterated” standards, i.e. tri-deuterated T and ET (otherwise written as 3dT and 3dET).

Ap.B.3.The purpose of these standards is to measure variations in the efficiency of steroid extraction between different samples (and different external standards). For example, we could have two urine samples (A and B), one of which is relatively clear (urine A) and one of which is relatively cloudy (urine B). The cloudy B sample may contain dead skin cells or other substances which will interfere with steroid extraction. A greater percentage of the steroids will therefore be extracted from the clearer urine A than the cloudy urine B. These differences can be significant with a 20% variation between urine samples not unusual. The internal standards are used to correct for these variations in a relatively simple process. Since you know how much internal standard you have added to the sample, you can compare this with how much you ultimately extract. This will give you a relatively precise measure of your extraction efficiency of the internal standard. It is then reasonable to assume that the extraction efficiency of the non-deuterated endogenous steroid is the same as that for the internal standard. Simple arithmetic can then be used to correct any variations in extraction efficiency.

External Standards

Ap.B.4. External standards are standards that are run separately, or external, from the urine samples in question. These standards are used to enable quantification and identification of the compounds in the urine.

Ap.B.5. Identification of a compound is critical in any doping case. Although the procedure is relatively simple, its implementation is often poorly handled. In simple terms, if you wish to identify T and ET within a urine sample you must run a T and ET standard contemporaneously and on the same MS as the urine. The urinary T / ET and standard T / ET must share two traits in common for positive identification:

Similar GC retention times.

Identical Mass Spectrums.

(If you are not clear what these terms mean, please refer to Appendix A.)

Ap.B.6. As stated by the IOC medical code:

“…all the information collected …must be… in agreement with the known facts and structure of the doping agent or metabolite(s)”. (IOC Medical Code, Appendix D, Section 2.1.4.).

Ap.B.7. You will have noted in this report that at best only one of these indicators appears to have been used by the DCC in Mr. Edwards’ case.

Ap.B.8. Quantification is a more complex procedure and can be achieved in a number of different ways. Unfortunately, the DCC has refused to reveal the way in which they quantified T and ET in Mr. Edward’s case, quoting commercial confidentiality as the reason. It should be noted that the DCC has provided this information freely to other T cases! However, the standard method of quantification, and method used by the DCC in other T cases involves running a series of six or more standards containing a range of concentrations. To demonstrate this is a standard procedure it is useful to refer to EEC regulations on steroid analysis that state:

“A minimum of six calibration points is required, adequately distributed along the calibration curve.” Section 2.3.1.2, 93/256/EEC.

Ap.B.9. When these samples are passed through a mass spectrometer they will produce a range of response from the MS detector in amps. These can then be plotted against the known concentrations of the standards to give a calibration curve (see Figure 1, next page).

ApB.10. Once this calibration curve is produced, subsequent urine samples (the T and ET concentrations of which are unknown) can be quantified using the regression analysis shown in Figure 1. In other words, the MS will give you a value in amps. This should be read off the Y-axis of the graph and a line extended from this point until it intercepts the calibration curve. The value on the X-axis at this point gives the concentration of T in ng/ml.

ApB.11. Despite our best efforts over the past six years, this very basic analytical data has not been provided.

Figure 1. Idealized calibration curve run a GC-MS system which is perfectly linear. The curve has a perfect regression fit (r2 = 1) a slope of 1 and a Y intercept of 0.

Appendix C"