Diagnostic Uncertainty in the Interpretation of Small Atypical Acinar Lesions of Prostate To the Editor: The recently published article by Van der Kwast et al 1 presents interesting and important data on diagnosis of small atypical foci on prostate needle biopsies. The authors show that these small foci are not reliably diagnosed by community pathologists (k—0.21), with only slightly better performance by recog- nized experts in the field (k—0.39). Not surprisingly, the results are worse (ie, consensus is particularly evasive) for the smallest foci under considera- tion, consisting of 5 or fewer acini. The authors are to be commended for a carefully designed study with the participation of leading experts in prostate pathology. We write, how- ever, to take strenuous issue with the conclusion as stated in the abstract; the authors ‘‘encourage pathologists to obtain intercollegial consultation of a specialist pathologist for these lesions before a carcinoma diagno- sis.’’ Why? This conclusion is not supported by their data. Even if shown to an expert, the diagnosis would depend on which expert one chose as a consultant, and we have no way of knowing which consultant is right in a given case. We would interpret their results as follows: (1) small foci of 5 or fewer glands cannot be reproducibly diagnosed as benign versus malignant, even by experts, and therefore (2) major treatment decisions should not be made based on such small foci, given the lack of reproducibility of diagnosis. We see no benefit to the patient of obtaining an expert opinion in such a case. The underlying issue is how reproducible the diagnosis of a given histopathologic variable must be to use it to guide therapy. We are not aware of this being formally discussed or rigorously examined, yet it is critical. It is a highly complex issue, requiring consideration of the nature of the treatment options, but a few recent examples are available for consideration. In the case of adminis- tration of adjuvant therapy in breast cancer, American Society of Clinical Oncology/College of American Patho- logists guidelines call for test accuracy (agreement with independent refer- ence method) of 95% for HER2 testing. Frozen section has an accep- table accuracy of greater than 98%. The diagnosis of single small atypical acini on prostate biopsies falls well short of these standards. As we move to evidence-based treatment deci- sions, greater reproducibility will be expected of pathologists. We should be prepared to state when we are not able to render reproducible diagnoses, as our opinions remain critical in treatment planning, and our credibility is eroded if we pretend all diagnoses are equal, in terms of reproducibility. The authors have clearly identified a circumstance where a diagnosis of carcinoma is not possible with cer- tainty, an important step forward. We should now search for tools that will objectively predict patient out- come in such cases, and not fall back on reliance on subjective expert opi- nion for these small atypical acinar lesions. C. Blake Gilks, MD, FRCPC David G. Huntsman, MD Department of Pathology Vancouver General Hospital and British Columbia Cancer Agency Canada REFERENCE 1. Van der Kwast TH, Evans A, Lockwood G, et al. Variability in diagnostic opinion among pathologists for single small atypical foci in prostate biopsies. Am J Surg Pathol. 2010;34: 169–177. In the Diagnosis of Limited Prostate Cancer, Is Observer Variability an Important Consideration When Compared With Variability of Patient Outcome? To the Editor: I read with interest, the recent article regarding variability in diagnosis of small atypical foci in prostate biop- sies. 2 This paper was authored in large part by experts and the conclusion was that experts should be consulted in cases in which there are very few cancerous appearing acini present. The problem here is that experts come and go. The biology of prostate cancer is of greater stability. The variability of ex- pert opinion is not only a concern at a given time but also through time. This should draw attention to an article too often ignored, and perhaps now viewed as ancient. Henson 1 outlines well the principle of how variability in outcome should be compared with variability in opinion. It is dubious that in a setting of close surveillance, and given the bio- logic characteristics of prostate cancer, that variability in the ability of patho- logists to diagnose a few suspicious acini as cancer instead of atypical will significantly alter ultimate patient out- come. 3 There is not an insignificant morbidity associated in the treatment of prostate cancer, and our pathology literature typically ignores morbidity at the expense of K-M survival plots. Impotence and/or incontinence for a ‘‘vanishing cancer’’ or an ‘‘autopsy can- cer’’ are not desired outcomes that we would wish to be contributors. ‘‘First, do no harm,’’ comes to mind. Outlining a problem should not come without at least a suggestion for a solution. My suggestion for consi- deration is an extent criterium. I find informally that at least 10 malignant seeming acini can be diagnosed as cancer with relative ease within a ‘‘community’’ level setting of practice. Let me propose that if there are fewer than 10 possibly malignant acini, a diagnosis should default to suspicious/ atypical—no matter the credentials of LETTERS TO THE EDITOR Am J Surg Pathol  Volume 34, Number 7, July 2010 www.ajsp.com | 1071