Article Text

Polymerase chain reaction in the diagnosis of bacterial endophthalmitis
  1. Department of Clinical Ophthalmology, Institute of Ophthalmology, 11–43 Bath Street, London EC1V 9EL
    1. K L THERESE,
    2. A R ANAND,
    1. Vision Research Foundation, Sankara Nethralaya
    2. 18 College Road, Chennai-600 006, India

      Statistics from

      Editor,—The paper by Therese et al1 raises several issues which require clarification. The contamination of Taq polymerase by bacterial DNA is now well established in the published press. Taq DNA polymerase is known to be contaminated with low levels of bacterial DNA not originating from either Thermus aquaticus orEscherichia coli and is easily amplified using universal bacterial primers based on ribosomal gene sequences.2-4 Although this level of contamination is insufficient to give a detectable amplification product after just one round of polymerase chain reaction (PCR), it is easily detected following nested amplification. The specific Taq used in the study (AmpliTaq DNA polymerase) is well known for being unsuitable for bacterial PCR using pan-bacterial primers such that the company itself (Perkin-Elmer) has more recently introduced a “low DNA” Taq (Amplitaq LD) in order to reduce the size of the problem. The reduced level of contamination in this Taq is still sufficient to yield positive “negative controls” after two rounds of PCR with eubacterial primers. Therefore, before first round amplification, it is of paramount importance to pretreat the Taq polymerase with restriction enzymes (unpublished observations), and to include the first round negative control as a test sample in the nested PCR reaction. The levels of DNA contamination are easily detectable at the sensitivity (40 fg) for the second round PCR reported by this group of authors and neither in the text nor in the figures is there any mention of a first round negative control as a test sample in the nested PCR reaction. The results submitted by Professor Madhavan’s group reflect PCR in the absence of adequate negative controls and are, therefore, meaningless.

      It is also well known that 22–43% of anterior chamber cultures are positive immediately after cataract surgery in patients that subsequently do not develop endophthalmitis.5-7 Not only has no attempt been made to provide clinical data about the cases with endophthalmitis but also no information is provided about whether these samples were from cases of acute/chronic/delayed endophthalmitis cases. With the high sensitivity of PCR and the ability to detect non-viable organisms, a higher yield of positive results is only to be expected. But, for example, a positive PCR result in the absence of a positive culture result a few days postoperatively is not necessarily evidence of infection sufficient to cause endophthalmitis. Also, in the absence of speciation no information is obtained regarding the virulence of the organism. All “PCR” based techniques for investigation of cases of presumed bacterial endophthalmitis should, therefore, be accompanied by clinical data to allow readers to judge for themselves whether the results obtained are truly applicable to the clinical setting.

      The contamination-free method of collection of samples is always critical but especially so if the detection method involves PCR techniques. No details of the preoperative/presampling preparation method are provided and no information is given as to whether the procedure was standardised and how many surgeons were involved in the collection process.

      The only sensitivity data reported are from extracted dilutions of DNA and not from live organisms. As the ability to extract DNA from intact cells is an integral step in any DNA amplification method this is another major flaw in this study.

      It is not surprising that Madhavan et al had little success in culturing P acnes since the technique used was incorrect: cultures should be maintained for up to 14 days instead of only 10.8

      The statement that “PCR showed 100% correlation with smear and culture results” is erroneous and misleading as this can not be verified in the absence of speciation techniques to identify the PCR product/s amplified. The final paragraph begins “Further studies are needed to identify the specific eubacterial strains …”. The presence of different strains is irrelevant as treatment is identical. We suggest Madhavan’s group first attempt to identify the bacterial species present. Any new diagnostic test should be evaluated in terms of its clinicalspecificity as well as sensitivity: Therese et alhave not addressed the specificity of PCR in the detection of disease so no comments on its clinical usefulness are warranted. Their suggestion that “Hence, the anterior chamber tap could be the method of choice in the diagnosis of endophthalmitis when a highly sensitive technique such as PCR is applied” has no basis and is likely to lead to mismanagement. The anterior chamber is the site of entry of organisms in the majority of cases. The presence of a positive PCR does not always correlate with established infection and the presence of a variety of bacteria from the patients own eyelid flora is only to be expected. Also, mixed infections have been reported in the published press.



      Editor,—We are extremely thankful to Okhravi’s group for their critical comments on our article and would like respond to the comments and queries raised by them.

      Regarding the point on the contamination of Taq polymerase with bacterial DNA and the adequacy of negative controls, we certainly were fully aware of this problem when this project was undertaken and therefore sufficient care was taken in providing proper and adequate controls in each and every step in the PCR which we feel was quite evident in the article.

      We included two negative controls in each of the amplifications (as mentioned in the article)—a reagent control and a sample extraction control. The sample extraction control consisted of sterile Milli Q water subjected to the same extraction procedures as the specimens. The second round controls consisted of a reagent control (only reagents used for the PCR reaction) and a sample extraction control, where 1 μl of amplified product from the first round extraction control was added. Only when both the reagent and sample extraction controls were negative were results on specimens accepted. Whenever negative controls indicated contamination the results were rejected.

      We can authentically state that the AmpliTaq DNA polymerase (Perkin-Elmer) used in our study did not contain any detectable amount of bacterial DNA, under our PCR conditions. For meaningful interpretations as suggested by Okhravi et alof the PCR results, we have indeed mentioned in the text (under paragraph “PCR using universal primers” p 1079, line 10), 1 μl of amplified product of the first round was used for the second round. It should be understood that it also included the negative controls. Therefore, we submit that adequate negative controls were used along with each reaction. Another observation which strongly indicated that the positive findings do not represent contaminants was that a significant number of clinical specimens were negative.

      Regarding their comments on the clinical data provided on endophthalmitis cases included in our study and their objection to our statement that anterior chamber tap (AC tap) is the method of choice in the diagnosis of bacterial endophthalmitis when PCR is applied, we need to state the following: we believe the clinical data (acute/chronic/delayed endophthalmitis) as suggested by Okhraviet al was beyond the scope of this study, because most of our postoperative endophthalmitis cases were referred to our hospital several weeks after their surgery and the bacterial agents which might have normally entered the anterior chamber during the immediate postoperative period could not have interfered with the PCR results of AC tap, unless they themselves were the causative agents of endophthalmitis, when the PCR automatically became true positive. Therefore, our conclusion that PCR on AC tap could be the method of choice as a diagnostic technique in cases of suspected bacterial endophthalmitis is correct.

      We included clinically evident post traumatic and endogenous infective endophthalmitis cases in addition to postoperative ones to highlight the diagnostic value of PCR on AC tap in all these three clinical groups since the earlier study of Hykin et al  was based only on vitreous aspirates from delayed postoperative endophthalmitis cases.

      In response to their statement “In the absence of speciation, no information is obtained regarding the virulence of the organisms”, we wish to state that as our study was aimed only at evaluating the diagnostic value of PCR in bacterial endophthalmitis, speciation and virulence of bacteria with reference to the clinical data were irrelevant and did not need to be included in our study.

      Regarding preoperative/presampling preparation method used for collection of intraocular specimens included in our study, they were collected by surgeons who used well established preoperative and presampling preparation methods for such collections, be it for PCR or other purposes. Therefore, our statement that the specimens were collected “aseptically”, we felt, did not need further elaboration into details of these established procedures.

      The PCR sensitivity data reported in our article were only for DNA extracted from living strains of Staphylococcus epidermidis and Propionibacterium acnes. We believe it was understood in the statement we made.

      In response to their statement that cultures for P acnes should be “maintained for up to 14” days instead of only 10, we wish to state that in our several years of experience,P acnes, if viable, has been isolated within 5–6 days of the incubation period and extended incubation even up to 45 days in culture media did not result in isolation of this bacterium. As we did not find it useful to incubate the inoculated media any further than 10 days, the media were discarded if there was no growth. But we certainly appreciate the suggestion of Okhraviet al in this procedure.

      Our statement that nested PCR showed “100% correlation with” bacteriologically (smear and culture) positive specimens was made to emphasise the exquisite specificity of the PCR method to detect eubacterial genome and no attempts have been made to speciate the amplified product to correlate with any isolated bacterium. We have, however, proposed “to identify specific bacterial strains in the specimens positive for eubacterial genome but negative forP acnes genome”. But at the same time, Okhravi et al make a contradictory statement that “the presence of different strains is irrelevant, as treatment is identical”. Identity of the bacterium, we feel, is useful to help clinicians to decide on the method of treatment.


      Supplementary materials

      Related Data

      Request Permissions

      If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.