 |
 |
 |
|
| home | CME | help | feedback | subscriptions | archive |
|
© 1997 The Journal of Bone and Joint Surgery, Inc.
CorrespondenceTO THE EDITOR: We welcome the article "The Role of Intraoperative Frozen Sections in Revision Total Joint Arthroplasty" (77-A: 1807–1813, Dec. 1995), by Feldman et al., which demonstrated the diagnostic utility of histological study of intraoperative frozen sections for distinguishing septic from aseptic loosening of joint replacements. This technique has been in use for more than five years in Oxford, where more than 300 hip and knee arthroplasties have been investigated in this way. We reported our experience1 with the first 106 hips in 1995. In our study, the main histopathological criterion for the diagnosis of highly suspected infection was the presence of an average of one neutrophilic polymorphonuclear leukocyte (neutrophil) per high-power field after examination of at least ten high-power fields. This criterion differs substantially from that of Feldman et al. and others3, who recommended more than five neutrophils per high-power field. In our study, we found that, in at least two cases in which there was bacteriologically proved infection (as well as a strong clinical suspicion), histological study of frozen sections showed the presence of neutrophils but fewer than five per high-power field. We also noted that, in a number of cases of septic loosening, some samples of the tissue from around the replacement contained more than five neutrophils per high-power field, whereas others contained fewer than five. We strongly agree with Feldman et al. that careful sampling of the tissue surrounding a replacement is necessary because the nature and degree of the inflammatory response can vary markedly even within one specimen of tissue. Samples from most hips that have septic loosening contain a large number of neutrophils, but it should be appreciated that the presence of fewer of these cells does not exclude the possibility of this diagnosis. We1 also pointed out a number of other factors that should be taken into account in the histological assessment of frozen sections of tissue surrounding replacements. First, the pathologist should be informed of all relevant clinical details that could modify the inflammatory response in the tissue surrounding the replacement. In some cases of active rheumatoid disease, a heavy neutrophil infiltrate can be seen in the membrane; without previous knowledge of this condition, an erroneous diagnosis of septic loosening could be made on the basis of the frozen section. Previous treatment with antibiotics may also modify the nature of the inflammatory response, with more chronic inflammatory cells (particularly plasma cells) and fewer neutrophils present. In this regard, it should also be noted that a heavy infiltrate of plasma cells or the presence of numerous lymphoid aggregates can be seen in some cases of low-grade infection, in which the number of neutrophils may again be fewer than five per high-power field. Thus, although we agree that a diffuse acute inflammatory infiltrate is the most common finding of histological study of frozen sections of infected tissue from around replacements, it should be appreciated that this infiltrate may be quite focally distributed and may not be the predominant feature in the tissues examined. We1 recommended a combined clinical and pathological (both histopathological and microbiological) approach to the diagnosis of septic loosening, as only in this way can the appropriate operative treatment of the failed arthroplasty be ensured.
N. A. Athanasou, Ph.D., M.R.C.P., M.R.C. Path.; R. Pandey, M.S. (Orth), Dip. (Orth); R. de Steiger, F.R.A.C.S. (Orth); P. McLardy Smith, M.S., F.R.C.S.: Nuffield Orthopaedic Centre NHS Trust, Windmill Road, Headington, Oxford OX3 7LD, United Kingdom
Dr. Feldman, Dr. Lonner, Dr. Desai, and Dr. Zuckerman reply:
Markedly different histopathological criteria were used in the various studies of the utility of intraoperative frozen sections in revision total hip and knee arthroplasty. In the study by Athanasou et al.1, a frozen section was considered to demonstrate infection if an average of at least one polymorphonuclear leukocyte per high-power field was noted on examination of ten fields. We used the criteria previously described by Mirra et al.3—namely, the presence of at least five polymorphonuclear leukocytes per high-power field after at least five fields have been studied under forty times magnification. While we did not average the number of polymorphonuclear leukocytes per high-power field in at least ten fields, more than ten fields were routinely analyzed in almost every case; the five most cellular fields were considered in the interpretation of the specimen and in the reporting of the results. On the basis of the difference in interpretation of the histological data, greater sensitivity but less specificity would be expected with use of the criteria of Athanasou et al.1. However, neither was found. In fact, the specificity in both our series and their study was 96 per cent; there were no false-negative frozen sections in our retrospective series of thirty-three patients, for a sensitivity of 100 per cent compared with that of 90 per cent reported by Athanasou et al. Our pilot study was followed by a larger, prospective study in which we analyzed 175 consecutive revision total hip and total knee replacements2. With use of similar criteria for determination of infection, positive frozen sections were classified further as having either five to nine polymorphonuclear leukocytes per high-power field or ten polymorphonuclear leukocytes or more per high-power field. Of the 175 frozen sections examined, twenty-three demonstrated at least five polymorphonuclear leukocytes per high-power field; of these, five sections demonstrated five to nine polymorphonuclear leukocytes per high-power field, while the remaining eighteen had ten polymorphonuclear leukocytes or more per high-power field. Of these twenty-three cases, sixteen were considered to be infected on the basis of final cultures and permanent histological sections; all of these had ten polymorphonuclear leukocytes or more per high-power field, and none had five to nine polymorphonuclear leukocytes per high-power field on frozen section. The remaining 152 frozen sections were considered negative, exhibiting fewer than five polymorphonuclear leukocytes per high-power field. Of the 152 hips with a negative frozen section, three were considered to be infected on the basis of final cultures. These data yielded a sensitivity of 84 per cent, whether the criterion of five to nine polymorphonuclear leukocytes or ten polymorphonuclear leukocytes or more per high-power field was used. The specificity of intraoperative frozen section increased from 96 to 99 per cent, depending on whether the former or latter criterion was used. The positive predictive value of intraoperative frozen section was only 70 per cent if an index of five to nine polymorphonuclear leukocytes per high-power field was used compared with 89 per cent if ten polymorphonuclear leukocytes or more per high-power field was used as the index; this was a significant difference (p < 0.05). Frozen section was equally effective at predicting the absence of infection with the use of either index. The negative predictive value of each was 98 per cent. The statistical values calculated from our data were remarkably similar to those in the study by Athanasou et al.1. In our series, three negative frozen sections with fewer than five polymorphonuclear leukocytes per high-power field were obtained from infected hips, accounting for the sensitivity of 84 per cent. While this may not be significantly different from the sensitivity of 90 per cent found with use of the criteria of Athanasou et al., frozen section may have proved to be a more sensitive test if we had used their criteria. Perhaps, then, the likelihood of frozen section predicting infection is identical if the polymorphonuclear leukocyte count is averaged over ten high-power fields with use of their criterion or over five high-power fields with use of our criterion. While we would expect a large number of false-positive frozen sections if fewer than ten polymorphonuclear leukocytes per high-power field were used as the index, the criterion of Athanasou et al., as reported, does not lack adequate specificity. The question remains: when using the average number of polymorphonuclear leukocytes per high-power field present in ten high-power fields as the criterion, did Athanasou et al. consider equally those cases with ten polymorphonuclear leukocytes in one high-power field and those with one polymorphonuclear leukocyte in each of ten high-power fields? Despite their results, we believe that frozen section is truly suggestive of infection only if a high number of polymorphonuclear leukocytes per high-power field is noted in at least five of the high-power fields analyzed. We do have several concerns regarding the study by Athanasou et al.1.
1. Their study was retrospective. Were the pathology slides reviewed again to confirm their findings? What was the clinical follow-up? Was the clinical outcome consistent with that predicted by histopathological analysis? 2. One case had frozen-section data for two separate revision procedures. Was the second procedure a reimplantation after resection arthroplasty? Were all frozen sections representative of only the initial revision procedure or did some include tissue from the second-stage procedure? We have found that frozen section at the time of second-stage reimplantation (particularly if it is within two months) is highly unreliable as a large number of acute inflammatory cells may be present consistent with healing of the operative incision. 3. Unlike Athanasou et al., we as well as Mirra et al.3 did not find a noteworthy association between chronic inflammatory response and infection.
In conclusion, although we agree that acute inflammatory response may be quite focal in the face of infection around a total joint replacement, we do not necessarily agree with the criterion of Athanasou et al.1. Admittedly, our data were reported and calculated differently than theirs. Certainly, sampling error may be an ever-present problem, but in appropriately obtained and analyzed tissue this error should be minimized. The pathologist must be well versed in the technique and interpretation of frozen section for this truly to be considered a reliable test of acute infection around a joint. Additionally, the orthopaedic surgeon and the pathologist must have a close working relationship and free exchange of information to optimize the care of the patient in this setting.
David S. Feldman, M.D.; Jess H. Lonner, M.D.; Panna Desai, M.D.; Joseph D. Zuckerman, M.D.: Departments of Orthopaedic Surgery (D. S. F., J. H. L., and J. D. Z.) and Pathology (P. D.), Hospital for Joint Diseases, 301 East 17th Street, New York, N.Y. 10003
References
  CiteULike  Connotea  Del.icio.us  Facebook  Technorati  Twitter What's this?
This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| home | CME | help | feedback | subscriptions | archive | site index |
|
Stanford University Libraries' HighWire Press (TM) assists in the publication of JBJS Online. |