Serum samples in clinical study were manipulatedBMJ 1998; 316 doi: http://dx.doi.org/10.1136/bmj.316.7131.628 (Published 14 February 1998) Cite this as: BMJ 1998;316:628
- R Szibor, Professor of natural sciences,
- W Kuchheuser, Physician,
- C Luley, Professor of medicine,
- D Krause, Professor of forensic medicine
EDITOR—Smith has announced that a Committee on Publication Ethics had been formed as a response to the increase in misconduct in research, as, for example, in the case reported by Dyer in the same issue. 1 2 Pharmaceutical companies should be anxious to detect and disclose any cases of research fraud because they may affect the credibility of the companies' clinical drug studies. We present here an example to show that methods are available for this purpose. A pharmaceutical manufacturer asked us to investigate the case, which turned out to be one of serious fraud.
Some measurements made on serum samples sent in by an investigating doctor had been questioned in a central laboratory. We were asked to check the identity of these 10 pairs of serum samples (1a and 1b, …, 10a and 10b). The test protocol stated that each pair of samples came from one patient.
After high speed centrifugation we obtained enough cells to determine the DNA markers amelogenin,3 humTH01, humACTBP2, and humFIBRA (microsatellites survey).4 The supernatant was used to investigate the serum markers GC, AHSG, HP, and TF with isoelectric focusing.5 In none of the 10 cases could the samples a and b be assigned to the same patient. In four cases a change of sex was noted (amelogenin XX→XY). Moreover, all serum samples showed three or four alleles in one or several markers so that pooled serum from at least two people was present. In four cases the serum came from at least three people. To provide an example, the table gives the results for samples 5a and 5b and 6a and 6b. There is every reason to believe that these were intentional manipulations.
With the great number of clinical studies being performed, it is difficult to find out whether there are other undetected, though less evident, cases of this kind. In the present case, an irregularity was assumed owing to the extremely improbable findings in the determination of relevant laboratory variables.
To prevent cases such as this one, random identity controls should be established by means of forensic markers as part of clinical drug studies, before publication. Thus potential misconduct would be deterred. The use of the above mentioned serum markers in combination with three or four DNA markers allows detection of almost all irregularities of the kind described in this case.
Log in using your username and password
Log in through your institution
Register for a free trial to thebmj.com to receive unlimited access to all content on thebmj.com for 14 days.
Sign up for a free trial