Immunohistochemistry for basal cell-specific and cancer-associated markers is valuable in the histopathological diagnosis of prostate carcinoma in needle biopsies [2, 3]. For example, positive AMACR staining helps identify areas of carcinoma, whilst high molecular-weight CKs and the p63 transcription factor aid the search for normal basal cells that are typically absent in malignant prostatic glands. However, using single markers to diagnose prostate carcinoma is of limited use. For example, AMACR may also be expressed in high grade PIN, in atypical adenomatous hyperplasia (adenosis) and even in atrophic or benign glands [3, 7]. Moreover, the distribution of basal cells can be patchy in both normal glands and in some benign lesions that mimic prostate cancer, such as areas of atrophy, post-atrophic hyperplasia, and atypical adenomatous hyperplasia (adenosis) . Furthermore, a diagnosis that is reliant on the absence of an immunohistochemical reaction is inherently less reliable. Indeed, a variety of technical problems, including excessive formalin fixation , may lead to false negative staining of normal basal cells.
The importance of these immunohistochemical and technical pitfalls in interfering with accurate diagnosis can be reduced by combined staining of prostate biopsies for both positive and negative cancer markers using antibody cocktails. Typically, these cocktails include AMACR together with either an antibody to high molecule-weight CK or to p63. Study by Trpkov et al. demonstrated that CK5/6 is an excellent and dependable basal cell marker when used in combination with AMACR; and CK 5/6 exhibited excellent specificity for prostate cancer, which uniformly lacked CK5/6 staining  . Some workers have gone further by including an additional marker for basal cells in three-marker cocktails, in order to try to improve the precision of prostate carcinoma diagnosis in limited biopsy material. Thus, Jiang and co-workers used immunohistochemistry with a triple-antibody cocktail (containing antibodies to AMACR, high molecule-weight 34βE12, and p63) to identify small, focal prostate carcinomas with high sensitivity and complete specificity . Similarly, Ng and colleagues used the same triple-cocktail in a tissue microarray study to identify prostate carcinoma with improved sensitivity (93.8%) and specificity (100%), compared with using the three antibodies individually .
Our study supports these previous findings. In addition, we show for the first time that antibody to the high molecular-weight CK5 can be successfully incorporated in triple cocktails together with AMACR and p63. CK5 is widely used as a robust immunohistochemical marker for, amongst other things, basal epithelial cells. Thus, antibodies to CK5 are widely available as routine markers in diagnostic pathology laboratories, and can easily be included in immunohistochemical protocols using automatic staining machines. Although high molecular-weight 34βE12 is included in antibody cocktails for diagnosing prostate carcinoma, there are potential drawbacks associated with the use of this marker. For example, 34βE12 antibody reacts with a wide variety of CKs including not only CK1, CK5, CK10 and CK14, but also an undefined CK that may stain at least some complex epithelia. Thus, 34βE12 may stain breast secretory cells and ductal carcinoma in situ of the breast (which are negative with CK5 antibody), hampering the use of the antibody for the identification of CK5 and myoepithelial cells in the breast . Although this type of reaction has not been reported as occurring in the prostate, any degree of uncertainty about the specificity of a diagnostic antibody is worrying. Similarly p63 when used individually demonstrated loss of immunostaining intensity in stored slides and showed spurious cytoplasmic staining of the luminal cells when used in low dilutions .
In our study, similar kappa values were found for both the two-marker and the three-marker cocktails when used by both pathologists, indicating fairly good agreement between them for both cocktails. Both pathologists were equally comfortable evaluating the results of both antibody combinations. In the case of the three-marker cocktail, usage of red staining for CK5 and brown staining for both p63 and AMACR proved to be technically feasible as well as provided a good contrast for interpretation.
Both observers had similar sensitivity levels for correctly assigning biopsies to the diagnostic groups. Comparing use of the three-antibody with the two-antibody cocktail, diagnostic sensitivity was marginally improved for the junior pathologist (76.4% vs 75.7%, respectively), but was slightly lower for the senior pathologist (66.6% vs 77.4%, respectively). Although the explanation for this is not clear, one likely possibility is that the senior pathologist’s long experience with interpreting stains produced with two-marker cocktails resulted in a paradoxical loss of sensitivity when switching to a novel immunohistochemical stain. Nonetheless, both pathologists showed a marked improvement in the specificity of their various diagnoses comparing the three-antibody with the two-antibody cocktails (90.3% vs 71.8% specificity, respectively, for the junior pathologist and 90.3% vs 68.7% specificity, respectively, for the senior pathologist). The reasons for this lower false negativity for the basal cell markers in our study could be derived from previous studies stating that CK 5/6 is preserved in the basal cells of non-malignant glands that are cauterized, crushed or distorted for various reasons including laboratory procedures, testifying it’s robustness and reliability . These findings are also comparable to the study by Ng et al.; where 34βE12 was used and CK 5 was one of the several high molecular weight keratins detected by antibody to 34βE12 .