Correspondence

P-Glycoprotein and Multidrug Resistance

N Engl J Med 1996; 334:1335-1337May 16, 1996

Article

To the Editor:

Although the publication of questionable immunocytochemical illustrations — and results — is unfortunately not infrequent, Figure 1 of the article by Baldini et al. (Nov. 23 issue)1 cannot be ignored, particularly because the conclusions and suggested therapeutic implications of this study were based exclusively on the immunocytochemical data. Neither I nor 21 colleagues (including pathologists, nonpathologists, and Ph.D.s) could distinguish or differentiate the results shown in Figure 1B from those in Figure 1C, nor could we discriminate with any level of confidence the specific localization of antigen within the diffuse, fairly homogeneous, light-brown stained areas in Figure 1D. And since it is usually safe to assume that the best examples of laboratory results are those selected for publication, we do not understand how three independent observers scoring results such as these in 92 cases (with no reported interobserver variation) could have compiled consistent data.

Although the interpretation of immunocytochemical results in a histopathological context may be a complex matter, there should be nothing subtle or complicated about the qualitative evaluation or scoring of well-prepared, well-controlled immunoperoxidase slides. The dark brown diaminobenzidine precipitate should be discretely localized, with variations in intensity that reflect the concentration of antigen, sharply contrasting with the surrounding negative structures. Such contrast is simply not evident in these three panels of Figure 1. Two possible explanations (among others) are the following. First, the counterstain, hematoxylin, whose active dye form, hematein, has an absorption maximum (445 nm) that falls within the spectral range of the brown dyes, tends to blend in rather than contrast with the brown diaminobenzidine reaction product. Second, various degrees of diffuse background stain (partly masked by the counterstain in Figure 1C, but clearly evident in Figure 1D) appear as a diffuse saturation of brown hue in the photomicrographs.

Although clear and conclusive color photomicrographs of hematoxylin-counterstained immunoperoxidase localizations are quite possible, greater contrast and better appreciation of the background component can be achieved with dyes at the blue-green end of the spectrum, which are gaining acceptance as counterstains. They are sufficiently distinct from the brown diaminobenzidine reaction product to allow the eye — and the camera — to define the distribution of stain clearly, while providing adequate histologic detail. For example, immunoperoxidase reactions counterstained with methylene blue (absorption maximum, 663 to 667 nm) can be illustrated effectively by halftone photomicrographs shot with a blue filter that attenuates the counterstain while enhancing the brown precipitate2 — a separation not possible with hematoxylin.

No counterstain, however, can rectify ambiguous results, any more than it can relieve authors — and reviewers — of the charge to maintain uncompromising critical standards in publishing illustrations of key results.

Ricardo Mesa-Tejada, M.D.
Corning Clinical Laboratories, Teterboro, NJ 07608

2 References

1. 1

   Baldini N, Scotlandi K, Barbanti-Brodano G, et al. Expression of P-glycoprotein in high-grade osteosarcomas in relation to clinical outcome. N Engl J Med 1995;333:1380-1385
   Full Text | Web of Science | Medline

2. 2

   Mesa-Tejada R, Keydar I, Ramanarayanan M, Ohno T, Fenoglio C, Spiegelman S. Immunohistochemical detection of a cross-reacting virus antigen in mouse mammary tumors and human breast carcinomas. J Histochem Cytochem 1978;26:532-541
   CrossRef | Web of Science | Medline

To the Editor:

In their study of 92 patients treated for osteosarcoma, Baldini et al. found that “the presence of increased levels of P-glycoprotein in tumor cells is associated with a significantly increased risk of adverse events.” Among the 64 tumors that were considered negative for expression of P-glycoprotein, 56 showed no immunoreactivity for P-glycoprotein and 8 “had only scattered positive cells.” Since a tumor recurrence probably reflects the late expansion of only a subpopulation of the original tumor cells, it seems that these eight tumors may just as reasonably be categorized as positive. What are the data on the event-free survival of these eight patients? In a similar vein, were any of the recurrent tumors assayed for expression of P-glycoprotein?

J. Martin Johnston, M.D.
University of Wisconsin Children's Hospital, Madison, WI 53792-4672

To the Editor:

Baldini et al. reported a significantly increased rate of relapse in patients with osteosarcoma whose primary-tumor-biopsy specimen showed increased levels of P-glycoprotein. We found a decreased rate of spontaneous progression-free survival after nephrectomy in patients with renal-cell carcinoma whose primary-tumor specimens had increased expression of P-glycoprotein (as detected by the monoclonal antibody C219).1 These observations may support the suggestion by Pinedo and Giaccone in the same issue of the Journal that P-glycoprotein is a marker not only for chemoresistance, but also for increased tumor aggressiveness.2,3 An altered expression of adhesion molecules may enable tumor cells to invade the surrounding stroma and colonize distant organs. The VLA-2 collagen receptor (α₂β₁ integrin) is an adhesion molecule that can increase the frequency and distribution of metastatic lesions in an animal model.4

We compared multidrug-resistant with nonresistant renal-carcinoma cell lines with respect to the expression of VLA-2; VLA-2 was detectable exclusively in the multidrug-resistant (P-glycoprotein–positive) subline. The binding of multidrug-resistant tumor cells to type I collagen was increased and could be blocked by anti–VLA-2 monoclonal antibodies. Since multidrug resistance appears to be a multifactorial survival mechanism of tumor cells, several metabolic and phenotypic alterations may be expected with the induction of “classic” P-glycoprotein–mediated multidrug resistance. Whether an altered integrin phenotype in drug-resistant tumor cells occurs through the selection of subclones or by an effect on the regulation of cellular properties is unclear.

Stefan Duensing, M.D.
Arnold Ganser, M.D., Ph.D.
Jens Atzpodien, M.D.
Hannover University Medical Center, D-30623 Hannover, Germany

4 References

1. 1

   Duensing S, Dallmann I, Grosse J, et al. Immunocytochemical detection of P-glycoprotein: initial expression correlates with survival in renal cell carcinoma patients. Oncology 1994;51:309-313
   CrossRef | Web of Science | Medline

2. 2

   Pinedo HM, Giaccone G. P-glycoprotein -- a marker of cancer-cell behavior. N Engl J Med 1995;333:1417-1419
   Full Text | Web of Science | Medline

3. 3

   Weinstein RS, Jakate SM, Dominguez JM, et al. Relationship of the expression of the multidrug resistance gene product (P-glycoprotein) in human colon carcinoma to local tumor aggressiveness and lymph node metastasis. Cancer Res 1991;51:2720-2726
   Web of Science | Medline

4. 4

   Chan BM, Matsuura N, Takada Y, Zetter BR, Hemler ME. In vitro and in vivo consequences of VLA-2 expression on rhabdomyosarcoma cells. Science 1991;251:1600-1602
   CrossRef | Web of Science | Medline

Author/Editor Response

The authors reply:

To the Editor: We thank Dr. Mesa-Tejada for reminding us of the importance of image quality in enhancing scientific communication. This has little to do with the consistency of the underlying data, however. As a matter of fact, images of histologic material always lose some information and may often acquire new features in the transition from the original slide to the print and finally to the page of a journal. We certainly regret that this occurred in the case of the images in our paper, in which subtle differences between levels of P-glycoprotein immunoreactivity are not readily apparent and tumor osteoid may appear to resemble the background to observers not particularly familiar with bone disease. The new images shown here as Figure 1AFigure 1Anti–P-Glycoprotein Avidin–Biotin–Peroxidase Immunohistochemical Analysis with Diaminobenzidine Used as a Chromogen and Gill's Hematoxylin Used as a Counterstain., Figure 1B, Figure 1C, Figure 1D, Figure 1E should demonstrate our findings unequivocally.

Nevertheless, in the text of our article, the careful reader will find all the methodologic information needed to interpret our study correctly, including a description of the use of adequate controls. In our view, such information is at least as important to the value of a scientific report as the quality of reproduction of paradigmatic results. In any case, we partly disagree with Dr. Mesa-Tejada's technical comments and still prefer hematoxylin as a nuclear counterstain in immunohistochemical analysis, in accordance with general opinion.1,2

Dr. Johnston raises the important question of the proper criteria for evaluating the immunoreactivity of P-glycoprotein. This is a critical matter because of the variety of methods authors have followed. In agreement with the approach of Salmon et al.3 and Chan et al.,4 among others, we did not consider the presence of a few scattered positive cells to qualify a tumor for inclusion in the group of positive cases. In sarcomas, such cells are often reactive elements, such as macrophages, that are known to be positive for P-glycoprotein. Moreover, the artifactual precipitation of chromogens is not uncommon in these tissues, which are so rich in mineralized matrix. However, if the eight tumors in our series that had scattered positivity are considered separately, the risk of adverse events in that group is not significantly higher than in the tumors that are entirely negative for P-glycoprotein. We did not evaluate the expression of P-glycoprotein in tumors from patients with relapses, but we agree that the relation between multidrug resistance and metastatic behavior is a very important problem.5 In this respect, we appreciate the information provided by Duensing et al. We also found increased expression of the α₂β₁ integrin in multidrug-resistant human osteosarcoma cell lines.

Nicola Baldini, M.D.
Franco Bertoni, M.D.
Istituti Ortopedici Rizzoli, 40136 Bologna, Italy

5 References

1. 1

   Appendices. In: Taylor CR. Immunomicroscopy: a diagnostic tool for the surgical pathologist. Vol. 19 of Major problems in pathology. Philadelphia: W.B. Saunders, 1986:365-78.
   

2. 2

   Reisner HM, Wick MR. Theoretical and technical considerations for the use of monoclonal antibodies in diagnostic immunohistochemistry. In: Wick MR, Siegal GP, eds. Monoclonal antibodies in diagnostic immunohistochemistry. New York: Marcel Dekker, 1988:1-49.
   

3. 3

   Salmon SE, Grogan TM, Miller T, Scheper R, Dalton WS. Prediction of doxorubicin resistance in vitro in myeloma, lymphoma, and breast cancer by P-glycoprotein staining. J Natl Cancer Inst 1989;81:696-701
   CrossRef | Web of Science | Medline

4. 4

   Chan HSL, Haddad G, Thorner PS, et al. P-glycoprotein expression as a predictor of the outcome of therapy for neuroblastoma. N Engl J Med 1991;325:1608-1614
   Full Text | Web of Science | Medline

5. 5

   Biedler JL, Spengler BA. Reverse transformation of multidrug-resistant cells. Cancer Metastasis Rev 1994;13:191-207
   CrossRef | Web of Science | Medline