Distinguishing between ECAs and EMAs before planning the patient treatment is clinically important. When the tumor involves both the uterine endometrium and the endocervix, it becomes difficult to distinguish the primary site of the tumor during preoperative assessment of the limited sizes of biopsy or curetting specimens. In this study, we evaluate various p16INK4a expression patterns in both ECAs and EMAs. We also investigate the most appropriate and effective p16INK4a IHC scoring methods in distinguishing these two types of gynecologic cancers in Taiwanese women. Our valuable domestic data can be extrapolated to women in general and will be helpful in referring and managing such cases worldwide.
The p16INK4a (cyclin-dependent kinase inhibitor 4) is a tumor suppressor protein that binds to cyclin-cdk4/6 complexes, which blocks kinase activity and inhibits progression to the S phase of the cell cycle in the nucleus.[10, 15, 27–35] However, interpretation of IHC data of p16INK4a staining results is complicated because of its unclear biological significance of cytoplasmic staining and lack of universal accepted algorithm in scoring methodology. Cytoplasmic reactivity is often regarded as unexpected, unspecific event. Some consider only nucleic p16INK4a labeling in tumor cells to be positive and ignore cytoplasmic staining.[16, 32] Others state that both nucleic and cytoplasmic immunoreactivities in tumor cells are characteristic and are indeed due to p16INK4a expression. [24–28] It has also been reported that strong cytoplasmic staining in mammary carcinomas is associated with negative prognostic factors, such as low differentiation, p53, Ki-67 labeling etc. We have learned that despite nucleic expression, p53 tumor suppressor protein is localized on cell cytoplasm, where it is regarded as a way of functional inactivation. [29–31] From our data, we cannot draw any conclusion yet about the biological significance of cytoplasmic p16INK4a expression. The knowledge about the functional meaning of cytoplasmic p16INK4a expression is still limited and further large-scale studies are encouraged in various human tissues and tumors.
There are a variety of IHC scoring methods including computer-based plans in literatures, and still seems to be no generally accepted protocols in research laboratories and clinical practices for rating and scoring the immunostaining results. Comparing commercially derived computer-based programs with the conventional analyses by pathologists, there are still lacks of optimized and standardized IHC scoring algorithms. As a result, the objective accuracy did not significantly improve clinical outcome measures. [33–36]
There are also various quantitative scoring mechanisms of p16INK4a expression using various cut-off thresholds in literature. Without mentioning the grading of intensity, Vallmanya Llena FR reports the cut-off point for p16INK4a expression to be 15% positively staining extent. Fregonesi PA defines the cut-off point for p16INK4a expression to be 5% cells stained positively.  Khoury T used the positive staining area >50% as a cut-off. They all took both nucleic and cytoplasmic p16INK4a IHC staining into considerations. However, Huang HY regarded any nucleic labeling of p16INK4a to be positive, irrespective of cytoplasmic staining. Kommoss S only used the nucleic staining patterns for p16INK4a evaluation. Milde-Langosch K defined the 12-tier scoring system which was also used in this study. In addition, we investigated the three p16INK4a IHC scoring mechanisms and determined the most effective means in the distinction between ECAs and EMAs. These results can potentially be applied to future clinically diagnostic techniques, when using p16INK4aimmunohistochemistry.
McCluggage WG (2003) and Mittal K (2007) stated that a diffuse, strong staining pattern of p16INK4a, involving nearly all tumor cells tends to be an ECA, whereas, focal, patchy staining pattern of p16INK4a involving 0–50% of cells tends to be an EMA in routine whole-sectioned tissue slides.[15, 17] We did not use the patchy or diffuse pattern of p16INK4a IHC staining as a diagnostically distinctive criterion between ECAs and EMAs in this TMA study, because we think that cases with primary EMA may seem to over-express p16INK4a beyond the limited 1.5 mm core area and therefore mimic a diffuse pattern of ECA primary. Instead, we preferred to use the semi-quantitative scoring system in considering the 0–3 points of staining intensity and 0–4 points of staining area extent by multiplying both, yielding a range of score 0 to 12 points. We then divided the results by an appropriate cut-off threshold of 4 to a two-tier of negative (0–3 points) or positive (4–12 points) for interpretation. The mixed cytoplasmic and nucleic stains with varying degrees of intensity and extent in the same tissue section were not uncommon. These discrepancies of p16INK4a expression in different subcellular compartments (cytoplasmic vs. nucleic) may have caused significant difficulties in the scoring process.
In this study, we defined 3 scoring mechanisms of the p16INK4a IHC expressions, as follows: (1) independent cytoplasmic staining alone, irrespective of nucleic staining (Method C), (2) independent nucleic staining alone, irrespective of cytoplasmic staining (Method N), and (3) mixed cytoplasmic with nucleic expression, using mean of the sum of cytoplasmic score plus nucleic score (Method Mean of C plus N). Of the 14 ECA and 24 EMA tissue samples in this study, we found that only 2 (Method N as well as Method Mean of C plus N) out of the total 3 scoring methods showed significant frequency differences (p < 0.05), whereas the third scoring method (Method C) did not show a significant difference (p > 0.05) in distinguishing between ECAs and EMAs. (Table 1) We can not completely yet rule out the possibility of the indigenous heterogeneity within individual tumors, leading to different p16INK4a expression patterns in various areas within the same tissue samples, because of the limited number of cases (14 ECA and 24 EMA tissues) and limited core size (1.5 mm) of the tumor specimens in TMA. However, our data showed that cytoplasmic p16INK4a expression correlated significantly with nucleic p16INK4a expression (p = 0.007) in EMAs, but not to do so (p = 0.663) in ECA. In short, cytoplasmic and nucleic staining correlates closely in EMAs, but do not in ECAs.
For the p16INK4a-marker characteristics and test effectiveness of ECA and EMA discrimination, the goal is to minimize the chance or probability of false positive and false negative results, and to maximize the probability of true positive and true negative results. According to our data, one method based on C (Method C) does not show significant frequency difference in making distinction between ECAs and EMAs (p = 0.245). The sensitivity of Method C was 35.7%, indicating a high false negative rate, whereas, the specificity of Method C is 81.0%, indicating a favorable low false positive rate. Both the negative predictive value (69.0%) and the positive predictive value (55.6%) do not provide valuable information. However, the scoring of p16INK4a expression using the other 2 mechanisms, including Method N and Method Mean of C plus N, shows significant frequency differences in making distinction between ECAs and EMAs (p < 0.05). The highest sensitivity is 78.6% using Method N, the highest specificity is 85.7% using Method Mean of C plus N, the highest negative predictive value is 85% using Method N, whereas the highest positive predictive value is 77.0% using Method Mean of C plus N. Notably, Method Mean of C plus N has the highest overall accuracy (80%).
In summary, of the three p16INK4a-scoring mechanisms, Method N and Method Mean of C plus N are useful in distinguishing these two gynecologic adenocarcinomas (ECA vs. EMA), whereas Method C is not. Using the Method Mean of C plus N in p16INK4a-marker IHC assessments, deserves the most favorable test effectiveness and performance of all, which may not only assist physicians in making adequate diagnostic distinction between ECAs and EMAs, but also help individual patients by appropriate treatment options. Despite the finite number of cases, our data provides significant and valuable reference to verify that p16INK4a with appropriate scoring mechanisms can be applied in designing the appropriately diagnostic multi-marker panels in distinguishing between ECAs and EMAs.