Utility of five commonly used immunohistochemical markers TTF-1, Napsin A, CK7, CK5/6 and P63 in primary and metastatic adenocarcinoma and squamous cell carcinoma of the lung: a retrospective study of 246 fine needle aspiration cases

Recent advances in the targeted therapy of lung cancers require an accurate subclassification of NSCLC. FNA-based cytological diagnosis is often the crucial first step in the diagnosis of a lung mass [7,16,18]. FNA specimens have unique problems, including small amount of tumor cells within the specimen, and potential specimen preparation artifacts. These features could affect the diagnosis and subclassification of NSCLC, so it may be difficult to subclassify the tumor by using routine hematoxylin and eosin (H & E) stained sections. Therefore, IHC markers are frequently used to aid in the diagnosis. Recent publications have already addressed the issue in the subclassification of NSCLC via small tissue biopsies and the utility of IHC markers in the surgical pathology setting [24,27-29]. However, evidence-based studies are still necessary for the determination of the optimal utilization of IHC markers, particularly in FNA specimens. In this study, we have retrospectively studied five most commonly used IHC markers, including TTF-1, Napsin A, CK7, P63 and CK5/6 in the subclassification of NSCLC in a large cohort (n = 246 cases) of cytological FNA cases.

Of 246 FNA cases, 102 cases were primary NSCLC, including 72 ADCs and 30 SqCCs, and 144 cases were primary NSCLC metastases to other body sites. Of metastatic NSCLCs, there were 131 ADCs and 13 SqCCs. The patient clinical characteristics were summarized in the Table 2. There were no gender and age differences between the primary and metastatic NSCLC (P = 0.3638 and P = 0.4110, respectively). All cases of primary NSCLCs had a clinical presentation of lung masses and radiographic diagnosis/suspicion of lung carcinomas, and 38% of the cases (39/102) had subsequent surgical biopsy and/or resection. In metastatic cases, 47.2% of cases (68/144) had a history of NSCLCs. The location of metastatic tumors is summarized in the Table 3. The most frequent metastatic body sites were local lymph nodes (45.14%, 65/144 cases) and pleural cavity (29.86%, 43/144 cases), respectively.

From our experience, most small biopsies lung cancer cases can be classified based on morphological evaluation, while about 30% to 40% of cases need IHC study in order to accurately classify the tumor, particularly in metastatic carcinomas. The clinical separation of the two major subtypes (ADC vs SqCC) has significant impact on the targeted therapy. Thus, in this study we have focused on the evaluation of the clinical utility of five most commonly used IHC markers in the subclassification of ADC and SqCC.

The cytological features of SqCCs include pleomorphic large tumor cells with hyperchromatic nuclei, opaque or “hard” cytoplasm, intracytoplasmic processes, or other features characteristic of squamous differentiation (Figure 1). Immunostaining patterns of P63 and CK5/6 in SqCCs are shown in Figure 1 and Table 4. CK5/6 showed 100% positivity in both primary and metastatic tumors, whereas, P63 showed a 91.7% and 100% positivity in primary and metastatic tumors, respectively (Table 4). The sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of CK5/6 and P63 in primary and metastatic SqCC are summarized in Table 5. In primary SqCCs, CK5/6 showed a slightly higher sensitivity than P63 (100% vs 91.7%), and similar specificity as P63 (77.8% vs 78.3%). In metastatic SqCCs, both CK5/6 and P63 revealed to have the same sensitivity (100% vs 100%), but CK5/6 showed a higher specificity than that of P63 (84.6% vs 68.4%, P < 0.05).

In SqCCs, we also found that TTF-1 and Napsin A could stain entrapped bronchial epithelial cells (Figure 1D) and alveolar macrophages (data not shown) rather than tumor cells. In this circumstance, tumor cells were considered negative for TTF-1 and Napsin A.

In SqCCs, P63 and CK5/6 are commonly used markers. Human TP63 gene is located on the chromosome 3q27-29; and the expression of the gene produces the full-length protein P63 and the truncated protein P40 [22,24]. P63 can be detected in benign bronchial stem cells and in neoplastic cells with evidence of squamous differentiation [30]. CK5/6 is a high molecular weight cytokeratin and expressed in neoplasms of epithelial origin, including SqCC, mesothelial carcinoma and urothelial carcinoma [31]. P63 and CK5/6 have been used in the diagnosis of lung SqCC [23,24,27,32], however typically in much smaller cohorts (≤50 cases) [23,32] than our study here. Our data demonstrates that P63 and CK5/6 have the sensitivity and specificity of 91.7% and 78.3%, and 100% and 77.8%, respectively. We found that CK5/6 to be more sensitive in primary SqCC, and more specific for metastatic SqCC (Tables 3 and 4). Both of these markers however can be detected in subset of ADCs. For example, P63 expression can be seen in 21.7% of primary ADCs and 31.6% of metastatic ADCs (Table 4). However it usually shows weak and focal expression in contrast to strong and diffuse staining seen in SqCC. These observations arise the potential possibility of glandular differentiation in SqCC. It has also been reported that in cases of poorly differentiated ADCs, which show a decreased expression of Napsin A and TTF-1 [19,21], and a proportion of these tumors have been shown to be P63-positive [33,34]. Similarly, CK5/6 has also been detected in a subset of primary ADCs (22.2%) and metastatic ADCs (15.4%). In addition, CK5/6 has also been detected in some cases of pancreatic, endometrial and breast ADCs [31].

ΔNP63 (N-terminal-truncated protein isoform of TA63), the truncated form of P63 without the transactivation domain, can be identified by the antibody designated as P40. P40 has been reported to have better sensitivity and specificity than P63 in the identification of SqCCs. Our previous study, however, showed that in SqCC P40, P63 and CK5/6 had a sensitivity of 80.5%, 90.0% and 93.5% and a specificity of 80.0%, 89.6% and 80.0%, respectively [21].

In addition, we also found that CK7 was positive in 50% (7/14 cases) of primary SqCCs and 75% (3/4 cases) of metastatic SqCCs. However, our study only had few cases of SqCC. A further study with larger numbers is necessary to draw a conclusion.

The cytological features of ADCs include clusters of tumor cells with prominent nucleoli, predominate or overt mucin production, vacuolated cytoplasm, acinar formation and other features characteristic of glandular differentiation (Figure 2). Immunostaining patterns of TTF-1, Napsin A, and CK7 in ADCs are shown in Figure 2 and Table 4. TTF-1 showed 84.5% and 86.9% positivity in primary and metastatic ADCs, whereas, Napsin A showed 92% and 67.8% positivity in primary and metastatic ADC, and CK7 showed 93.8% and 100% positivity in primary and metastatic ADCs, respectively, (Table 4). The sensitivity, specificity, PPV and NPV of TTF-1, Napsin A and CK7 in primary and metastatic ADC are summarized in Table 6. TTF-1 showed similar sensitivity between primary and metastatic ADCs (84.5% vs 86.9%, P > 0.05), but a higher specificity in primary ADCs than metastatic ADCs (96.4% vs 87.5%, P < 0.05). Napsin A showed a higher sensitivity in primary ADC than that of metastatic ADCs (92.0% vs 67.8%, P < 0.05), and the same specificity in both primary and metastatic ADC (100% for both). The sensitivity and specificity of CK7 in primary and metastatic ADCs were 93.8% and 50%, and 100% and 25%. In addition to ADCs, we also found that CK7 showed 50% and 75% positivity in primary and metastatic SqCC, respectively. Taken together, TTF-1 had a better sensitivity, and Napsin A had a better specificity for the primary lung ADCs. Whereas, CK7 showed a suboptimal specificity for lung ADCs.

CK7, TTF-1 and Napsin A are the most commonly used primary lung ADC markers in daily practice. Although CK7 has been used for decades to identify lung ADCs, its suboptimal sensitivity and specificity are well-known [24,27,35]. TTF-1 is a nuclear transcript factor that is expressed in epithelial cells of the lung and thyroid. In the lung, it regulates the expression of genes involved in production of surfactant. The sensitivity and specificity of TTF-1 in the identification of lung origin vary, and range from 75% to over 95% [23,27,28,36]. However, TTF-1 is also immunoreactive in several other tumors, such as thyroid neoplasms, breast adenocarcinoma, gastrointestinal carcinomas, small cell lung carcinoma (SCLC), carcinoid and, possibly but controversially [37], primary lung squamous cell carcinoma [35,38-43]. Napsin A is a relatively new marker for primary lung ADCs [44]. It is a 35-kilodalton protein that is expressed in type II pneumocytes, alveolar macrophages, and renal tubular cells [44]. Functionally, it is an aspartic protease involved in the posttranslational modification of surfactant protein B (SP-B) in type II pneumocytes [45]. The expression of Napsin A has been shown to be transcriptionally regulated by TTF-1 [46]. Previous studies using surgical resected specimens indicated that Napsin A has a better sensitivity and specificity than TTF-1 in well to moderately differentiated lung ADCs [22,23,43]. Therefore, it has been used with TTF-1 together in the differential diagnosis of lung adenocarcinomas [23,47]. Napsin A may be particularly useful in poorly differentiated ADCs, which may lose TTF-1 expression [16,19].

In our study, CK7 has a sensitivity and specificity of 93.8% and 50.0% in primary lung ADCs, and 100% and 25.0% in metastatic lung ADCs. TTF-1 has the sensitivity and specificity of 84.5% and 96.4% in primary lung ADCs, and 86.9% and 87.5% in metastatic lung ADCs. Napsin A has sensitivity and specificity of 92.0% and 100% in the primary lung ADCs, and 67.8% and 100% in metastatic lung ADCs. Taken together, all three markers revealed similar sensitivities in primary lung ADCs; Napsin A showed the best and CK7 showed the worst specificity. In metastatic ADCs, CK7 showed better sensitivity than TTF-1 and Napsin A, but the worst specificity. In our medical institutions, Napsin A was introduced to the clinical practice in 2008. Interestingly, we found that the addition of Napsin A to routine practice coincided with an increase in the diagnosis of ADC subtypes from 14% to 36% and a concurrent decrease in NOS (otherwise not further classified) subtypes from 24% to 9% among NSCLC. This observation of improved subclassification of NSCLC with the use of Napsin A has also been reported elsewhere [22,23,36,48]. All studies, however, should also address limitations of Napsin A including: (a) Napsin A is positive in some cases of SCLC and lung carcinoid [42], (b) Napsin A is positive in some cases of renal cell carcinomas [43] and (c) Napsin A also stains pulmonary macrophages [37], which need to be distinguished morphologically from tumor cells. With these caveats in mind, we show that Napsin A exhibits strong specificity for ADCs of the lung origin.

We also examined the utility of these IHC markers in combinations. In SqCCs, we found that 14 cases of primary and 3 cases of metastatic tumors had both CK5/6 and P63 staining. Of SqCCs with paired IHC, we found that two P63 negative SqCCs were positive for CK5/6, and one CK5/6 negative SqCC was positive for P63. In primary ADCs, we found that 24 cases were stained with TTF-1 and Napsin A, 31 cases were stained with TTF-1 and CK7, and 14 cases were stained with TTF-1, Napsin A and CK7. In metastatic ADCs, we found that 9 cases were stained with TTF-1 and Napsin A, 52 cases were stained with TTF-1 and CK7, and 49 cases were stained with TTF-1, Napsin A and CK7. Of cases with paired IHC, six Napsin A negative cases were positive for TTF1, and two TTF-1 negative cases were positive for Napsin A.

To further test the sensitivity and specificity of IHC markers in combination, we used the bootstrap resampling approach to analyze both primary and metastatic NSCLCs. In ADCs, the utility of a combination of IHC markers (TTF-1, Napsin A and CK7) was compared with individual markers, with the results of primary ADCs shown in Figure 3, and metastatic ADCs shown in Figure 4. The sensitivity and specificity of combined markers were summarized in Table 7. Taken together, inclusion of TTF-1 and/or Napsin A in a panel shows an improved specificity in both primary and metastatic ADCs.

In SqCCs, the utility of combinations of IHC markers P63 and CK5/6 was compared with individual markers, and results of primary SqCCs are shown in Figure 5 and metastatic SqCCs are shown in Figure 6. The sensitivity and specificity of combined markers were summarized in Table 8. Although the specificity of P63 and CK5/6 is similar in primary SqCCs, CK5/6 shows better specificity in metastatic SqCCs.

The bootstrap method is a computer simulation technique based on random re-sampling of a dataset and subsequent analysis of the data distribution [26]. It is commonly used to estimate confidence intervals, the bias and variance of the actual dataset. Our data demonstrated that the combination of TTF-1 and Napsin A showed the highest specificity among three different combinations of TTF-1 + CK7, TTF1 + Napsin A and TTF-1 + CK7 + Napsin A in the identification of both primary and metastatic lung ADCs (Table 7). The specificity of P63 and CK5/6 is similar in primary SqCCs, CK5/6 shows better specificity in metastatic SqCCs (Table 8).

Based on the findings that TTF-1 and Napsin A have a high sensitivity and specificity for the diagnosis of primary lung ADCs, and CK5/6 stain is highly sensitive and specific for squamous differentiation, we outlined an algorithmic approach in the subclassification of NSCLC using FNA cases (Figure 7). In the algorithm, the evaluation of cytological morphology in the conjunction of immunostaining patterns is necessary for the final diagnosis of the tumor and/or further decision-making steps. For example: (a) ADCs should be favored for cases with both Napsin A and TTF-1 positivity; alternatively, either TTF-1 or Napsin A positivity, alongside CK5/6 negativity, (b) SqCCs should be favored for cases with CK5/6 positivity alongside Napsin A and TTF-1 negativity. The use of CK7 and P63 was not necessary to improve the identification of either ADC or SqCC. However, in poorly differentiated NSCLCs and/or adenosquamous carcinomas more IHC markers are needed. This approach may potentially improve the detection specificity, and may prove to be of diagnostic significance, as patients with poorly differentiated carcinomas may benefit from molecular screening for EGFR, KRAS or ALK mutations. As a note of caution, additional IHC markers are necessary to identify the origin of the tumor in cases of non-pulmonary carcinomas metastatic to the lung, and a careful clinical evaluation in these settings is also necessary [49]. The study by Su YC, et al., reported that CK7 was frequently detected in breast carcinoma in addition to lung ADCs, and CK20 was significantly more frequently detected in gastrointestinal carcinomas [50]. The other study by Kargi A, et al., reported that a subset of SCLC had a positive immunostaining pattern of TTF1 and negative immunostaining pattern of P63 [51]. Therefore, the interpretation of IHC patterns should be correlated with the morphological findings.