Authors issue a model retraction for mistaken bacterial identity

jcmA group in the Netherlands has retracted a case study on the diarrheal pathogen Campylobacter jejuni, commonly found in animal feces, after repeated tests showed the bacteria was actually C. fetus, which also causes spontaneous abortion in cows and sheep.

The 46-year-old man who had previously had an aortic valve replacement came to the doctors with endocarditis, an inflammation of the heart. Initial tests showed that it was due to a C. jejuni infection, which often lives in chickens, wombats, kangaroos, and sheep.

Only a few cases of endocarditis caused by C. jejuni had ever been reported. Unfortunately, a thorough followup made it clear that a different pathogen was at play. Let’s consider this retraction a model for all others in its clarity and thoroughness.

Take it away, notice for “Aortic Homograft Endocarditis Caused by Campylobacter jejuni“:

Volume 49, no. 11, p. 4016–4017, 2011. In our paper, we described the first case of homograft endocarditis caused by Campylobacter jejuni in a 46-year-old male. Only two other cases of C. jejuni endocarditis, involving native valves, had been reported before in the medical literature (A. Pönkä, T. Pitkänen, T. Pettersson, S. Aittoniemi, and T. U. Kosunen, Acta Med. Scand. 208:495–496, 1980; J. Torné Cachot, J. M. Garcés Jarque, R. Miralles Basseda, and A. García Flores, Rev. Clin. Esp. 184:114–115, 1989). The lightly curved, Gram-negative rods that grew from our patient’s blood culture were oxidase, catalase, hippurate, and indoxylacetate positive, corresponding with Campylobacter jejuni.

In our laboratory, a PCR-based screening method for diarrheal pathogens (T. Schuurman, R. F. de Boer, E. van Zanten, K. R. van Slochteren, H. R. Scheper, B. G. Dijk-Alberts, A. V. Möller, and A. M. Kooistra-Smid, J. Clin. Microbiol. 45:3692–3700, 2007) was recently evaluated for use in routine diagnostics, and the collection of strains used in this evaluation included the strain from the C. jejuni endocarditis patient. To our surprise, the result for the PCR assay for C. jejuni on this strain was negative. The strain was further analyzed by DNA sequencing of 16S ribosomal RNA, which showed 100% homology with Campylobacter fetus. Additional PCR assays specifically designed for the identification of C. fetus (C. Tramuta, D. Lacerenza, S. Zoppi, M. Goria, A. Dondo, E. Ferroglio, P. Nebbia, and S. Rosati, J. Vet. Diagn. Invest. 23:657–664, 2011; C. Abril, E. M. Vilei, I. Brodard, A. Burnens, J. Frey, and R. Miserez, Clin. Microbiol. Infect. 13:993–1000, 2007; S. Hum, K. Quinn, J. Brunner, and S. L. On, Aust. Vet. J. 75:827–831, 1997) gave positive results. An amplified fragment length polymorphism (B. Duim, P. A. Vandamme, A. Rigter, S. Laevens, J. R. Dijkstra, and J. A. Wagenaar, Microbiology 147:2729–2737, 2001) also indicated C. fetus. Therefore, we have to conclude that the patient from our case report did not suffer from an endocarditis caused by C. jejuni but instead from an endocarditis caused by C. fetus, which is a much more common phenomenon, as was also discussed in the case report.

We subsequently repeated the phenotypic tests on the patient’s strain. The result for the hippurate test was again positive, but the indoxylacetate test now gave a negative result. A positive hippurate test result would, however, already be sufficient to identify a campylobacter as C. jejuni (R. C. Jerris, P. I. Fields, and M. A. Nicholson, p., in L. S. Garcia and H. D. Isenberg, ed., Clinical Microbiology Procedures Handbook, vol. 1, 2010). In a Dutch report, a multiplex PCR assay for diarrheal pathogens was compared to phenotypic identifications (M. T. van der Beek, W. van Pelt, L. Heres, K. Veldman, J. A. Wagenaar, W. F. Jacobs-Reitsma, D. J. Mevius, and E. J. Kuiper, Dutch J. Med. Microbiol.,17:6–12, 2009). The authors of that report analyzed 1,585 isolates from 8 different laboratories that were phenotypically identified as C. jejuni and found that 96% of these identifications were correct (range, 81 to 100%). A hippurate test result may be false positive when the final color after addition of the 3.5%-ninhydrin solution is examined after the maximum incubation time. In our laboratory, the time between the addition of the reagent and the examination of the final color is strictly set at 10 minutes. However, with the data from the report by van der Beek et al. in mind, we decided to send the isolate to two other laboratories in The Netherlands, one of which was the WHO Collaborating Center for Campylobacter. In both laboratories, the result for the hippurate test was negative. A negative result for a PCR assay for hipO (V. Caner, Y. Cokal, C. Cetin, A. Sen, and N. Karagenc, Antonie van Leeuwenhoek 94:527–532, 2008) confirmed this. Results for additional phenotypic tests, including tests for growth at 25°C (growth), growth at 42°C (no growth), and susceptibility to nalidixic acid (resistant) and cephalothin (sensitive), also matched C. fetus.

Our case underlines in general the importance of thorough confirmation of unexpected bacterial species in clinically important cases, either by molecular methods or by extensive and reliable phenotypic tests. More specifically, it underscores the previously reported unreliability of the phenotypic hippurate hydrolysis test for the correct identification of Campylobacter spp. We hereby retract the paper and sincerely apologize for the inconvenience caused to the readers.

The paper has yet to be cited, according to Thomson Scientific’s Web of Knowledge.

It’s like we say in journalism: if your mother says she loves you, check it out, especially if you think she has wombat diarrhea bacteria.

3 thoughts on “Authors issue a model retraction for mistaken bacterial identity”

  1. Bravo! Not only is this an excellent explanation, it provides a valuable technical lesson: whenever possible, perform a stringent PCR test. A PCR screen for hipO early in the process would have avoided *all* the problems.

  2. I felt I learnt about C. jejuni and C fetus from retraction notice. Authors reiterate complexity of clinical sample testing for infectious agents as in this case. Nicely done and kudos to authors!

  3. There is a much more important and wider message here, to the publishers, I believe. Given the fact that many (most?) retraction notices to date have not been detailed, or very transparent about the background and even the details of the reasons for the retraction (e.g., lack of exact texts that were found to be identical in the cases of retractions resulting from “plagiarism”), and given the fact that COPE encourages precisely the level of transparency in a retraction notice that brings about clarity, as in this case, should publishers that have not published transparent and detailed retraction notices thus far be required to go back and rewrite opaque and incomplete retraction notices? It seems like we keep going back to square one on this topic i.e., who is actually responsible for the content of text written in the retraction notice, including the grammar, and tone? Does each publisher have a different policy and does each journal within a single publisher have a different policy? And if yes, how do such publishers reconcile different policies if they are equal COPE members (assuming that they are COPE members)?

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.