Immunohistochemical Expression of Neurotrophic Tyrosine Kinase Receptors 1 and 2 in Lung Carcinoma: Potential Discriminators Between Squamous and Nonsquamous Subtypes

Despite extensive study, there are few clinically useful immunohistochemical markers to help determine diagnosis and prognosis in lung carcinoma.1 This problem is particularly conspicuous in the differentiation between subtypes of non–small cell lung carcinoma, where correct identification of histologic subtype is essential to stratify patients to appropriate chemotherapeutic regimens.2 

The neurotrophic tyrosine kinase receptors NTRK1 (tropomyosin receptor kinase A [TRKA]) and NTRK2 (TRKB) belong to a family of nerve growth factor receptors that influence many aspects of neuronal development, including promoting proliferation and survival.3 In lung, NTRK signaling has been shown to promote cell survival and proliferation, whereas inhibition of NTRK signaling induces apoptosis and reduces clonogenicity.4–7 Potentially activating mutations in NTRK genes have also been found in subsets of large cell neuroendocrine carcinomas and lung adenocarcinomas, with the latter study finding a correlation between NTRK2 mutations and higher-stage disease.8,9 More recently, NTRK2 immunohistochemical expression in non–histologically subtyped non–small cell lung carcinoma has also been shown to correlate with lymph node metastases and higher-stage disease.10 

Immunohistochemical expression of NTRK1 and NTRK2 has been reported as strong in tissues exhibiting squamous differentiation but is less consistent or absent in glandular tissues,11 suggesting that these markers may be effective discriminators between squamous and glandular subtypes of non–small cell lung carcinoma. Immunohistochemical expression of NTRK1 and NTRK2 has been shown in lung cancer,4,10,12–14 but the expression profiles of these markers within lung carcinoma subtypes are unclear and the clinical utility of NTRK is unknown. Here we compare the immunohistochemical expression profiles of NTRK1 and NTRK2 in more than 680 unique cases of lung carcinoma of various histologic subtypes and correlate these profiles to patient outcome.

Normal, nonlung tumor, and lung cancer tissue samples, reviewed by pathologists with subspecialty expertise in lung pathology, were obtained from the archives of Vancouver General Hospital, St Paul's Hospital, and University of British Columbia Hospital (Vancouver, Canada) using ethics committee–approved protocols. The 08-011 lung cancer tissue microarray (TMA) has been previously described15 and contains 588 tandem 0.6-mm tissue cores from adenocarcinomas, squamous cell carcinomas, non–small cell carcinoma not otherwise specified, and large cell carcinomas collected between 1978 and 2002. The 08-012 lung cancer TMA is based on a previously described lung cancer TMA and contains tissue from 279 unique lung cancer cases collected between 1982 and 2001, including 78 squamous carcinomas, 61 classic carcinoids, 35 atypical carcinoids, 28 adenocarcinomas, 26 large cell carcinomas, 9 small cell carcinomas, 13 bronchioloalveolar carcinomas, and 6 large cell neuroendocrine tumors.16 Follow-up information is available for both lung cancer TMAs. Normal and malignant tissues in TMA format are used as immunohistochemical controls.

The monoclonal rabbit anti-human NTRK1 (clone 14G6, catalog no. 2508) and monoclonal rabbit anti-human NTRK2 (clone 80G2, catalog no. 4607) antibodies were purchased from Cell Signaling Technology (Beverly, Massachusetts). There is no reported antibody cross-reactivity of anti-NTRK1 with NTRK2 and anti-NTRK2 with NTRK1. Each TMA section was hybridized with either a 1∶350 dilution of anti-NTRK1 or a 1∶25 dilution of anti-NTRK2 using a Ventana automated immunostainer with standard CC1 heat-induced epitope retrieval protocol, Ultra-Rabbit horseradish peroxidase secondary antibody, and 3,3′-diaminobenzidine chromogen (Ventana, Tucson, Arizona). Immunostained slides were counterstained with hematoxylin. Parallel negative controls omitting primary antibody were also performed and did not show appreciable background staining. Sections of each TMA were also stained with hematoxylin and eosin using standard methods for histologic reference. The hematoxylin and eosin and immunostained TMA slides were scanned using a BLISS slide scanner (Bacus Laboratories, Lombard, Illinois) and are available for viewing at http://bliss.gpec.ubc.ca.

What constitutes positive NTRK1 and NTRK2 immunohistochemical staining is not well defined. Based on the biological function of these receptors and previous studies,10–12 we defined positive NTRK1 and NTRK2 immunostaining as circumferential membranous staining with or without cytoplasmic staining in one or more tumor cells regardless of staining intensity (Figure 1, A through D). Negative NTRK1 and NTRK2 immunostaining is defined as tumor cells exhibiting only nonmembranous staining (ie, only nuclear, nuclear and cytoplasmic, or only cytoplasmic; Figure 1, E), any staining in nontumor cells or cells not clearly recognizable as tumor (necrotic tissue, inflammatory cells; Figure 1, F), or an absence of staining (Figure 1, G and H). For the lung cancer arrays with tandem cores, the following rules were applied: when only 1 core was positive, the case was scored as positive; for cases with 1 uninterpretable core, the score of the interpretable core was used; when both cores were uninterpretable (ie, no tumor cells, no viable cells or folded/lost tissue core), the case was scored as uninterpretable.

To be included in the statistical analysis, each lung cancer case required at least 30 days of follow-up information and interpretable scores for both NTRK1 and NTRK2 immunostaining. Statistical analyses were done using SPSS v16.0 (Chicago, Illinois).

Prior to immunostaining the lung cancer TMAs, NTRK1 and NTRK2 immunostaining was assessed on control tissues. Although NTRK1 has been reported as weakly positive in the alveoli of normal lung,11 there was no staining of normal lung in the control TMA (0 of 1, Figure 2, A). NTRK1 expression has been reported as strong in normal epidermis, squamous mucosa, and epidermal carcinomas.11,13,17 Accordingly, NTRK1 staining was positive in normal epidermis (2 of 2), normal oral mucosa (1 of 1), basal cell carcinoma (1 of 1), and squamous carcinoma of skin (1 of 1), but absent in stomach (0 of 2) and colon carcinomas (0 of 4) in the control TMA (Figure 2, B, data not shown). NTRK2 has been reported as weakly expressed to undetectable in normal lung, liver, and heart,11 and similar results were obtained in the control tissue TMA (lung 0 of 1, liver 0 of 1, heart 0 of 1; Figure 2, C, data not shown). Unlike NTRK1, NTRK2 staining was undetectable in normal epidermis, normal oral mucosa, basal cell carcinoma, and squamous carcinoma of skin (Figure 2, D, data not shown).

One hundred eighty-one of the 867 individual lung cancer cases contained in the 08-011 and 08-012 TMAs are excluded from analysis because of uninterpretable cores or insufficient follow-up, leaving 686 cases for analysis. Patient characteristics for cases included in the analysis from TMA 08-011 are outlined in Table 1, and characteristics for cases included from TMA 08-012 are outlined in Table 2. The results of immunohistochemical staining for NTRK1 and NTRK2 by lung cancer subtype are summarized in Table 3. Representative images of immunostained tumors are shown in Figure 1, and the entire 08-011 and 08-012 TMAs immunostained for NTRK1, for NTRK2, and with hematoxylin-eosin are available for viewing at http://bliss.gpec.ubc.ca (last accessed May 31, 2010).

Positive NTRK1 expression correlates strongly to squamous carcinoma of lung (194 of 271 cases positive, Kendall τ-b  =  0.67, P < .001). Positive NTRK1 staining is also very specific (92.8%) but not very sensitive (71.6%) for squamous cell carcinoma compared with the other subtypes. Positive NTRK1 staining also occurs in the squamous portions of adenosquamous tumors (2 of 3), infrequently in large cell carcinomas (6 of 55), and rarely in adenocarcinomas (11 of 236) and carcinoid tumors (1 of 93). NTRK1 expression is absent in bronchioloalveolar carcinomas, pleomorphic carcinomas, large cell neuroendocrine tumors, and small cell carcinomas.

Expression of NTRK2 also correlates with squamous carcinoma of lung, although less so than NTRK1 (139 of 271 cases positive, Kendall τ-b  =  0.559, P < .001). Positive NTRK2 expression is highly specific for squamous carcinoma (96.4%) but is poorly sensitive (51.3%). Positive NTRK2 staining also occurs less frequently in large cell neuroendocrine tumors (2 of 6), adenosquamous carcinomas (1 of 3), and large cell carcinomas (4 of 55), and rarely in adenocarcinomas (11 of 236) and carcinoid tumors (1 of 93). Bronchioloalveolar carcinomas, pleomorphic carcinomas, and small cell carcinomas are consistently negative for NTRK2 expression.

Positive NTRK1 staining in squamous carcinoma has no prognostic significance with respect to disease-specific survival (DSS; log-rank test, χ² < .001, P > .99) or overall survival (OS; log-rank test, χ²  =  2.2, P  =  .14). Similarly, NTRK1 in adenocarcinoma has no prognostic significance in DSS (log-rank test, χ² < .001, P  =  .95) or OS (log-rank test, χ² < .001, P  =  .94). Positive NTRK1 staining is not frequent enough in other lung cancer subtypes to achieve statistically reliable correlations with patient outcome.

There is a significant positive correlation between positive NTRK2 staining in squamous carcinoma and improved DSS (log-rank test, χ²  =  11.8, P < .001; Figure 3, A) and OS (log-rank test, χ²  =  3.9, P  =  .047; Figure 3, B) in univariable analysis. The survival benefit of NTRK2 expression in squamous carcinoma remains significant in multivariable analysis including histologic grade and age for both DSS (Cox regression model: hazard ratio, 0.48; 95% confidence interval [CI], 0.33–0.70; P < .001) and OS (hazard ratio, 0.64, 95% CI; 0.48–0.87; P  =  .004). For a subset of cases (those in the 08-011 TMA), data on sex and smoking history are also available. NTRK2 continues to be an independent positive prognostic indicator for both DSS (hazard ratio, 0.40; 95% CI, 0.24–0.68; P < .001) and OS (hazard ratio, 0.59; 95% CI, 0.39–0.90; P  =  .01). Unfortunately, because of the wide time period during which the lung cancer specimens had been collected (1978–2002), accurate staging information is only available for the most recent cases, which precludes a reliable multivariable analysis to determine the relationship between NTRK2 expression and stage. We can, however, look at the correlation of NTRK2 expression with stage alone, where NTRK2 expression positively correlates with higher stage disease (Kendall τ-b, stage 1 versus stage 2/3  =  0.41, P  =  .03), indicating that the relationship between NTRK2 and improved outcome is not due to preferential expression in lower stage disease.

NTRK2 expression status in adenocarcinoma is not significantly correlated with DSS (log-rank test, χ²  =  1.3, P  =  .26) and OS (log-rank test, χ²  =  1.6, P  =  .21). Positive NTRK2 staining is not frequent enough in other lung cancer subtypes to achieve statistically reliable correlations with patient outcome.

NTRK1 and NTRK2 are preferentially expressed in squamous cell carcinoma of lung when compared with other histologic subtypes of lung carcinoma, most notably adenocarcinoma. This suggests that immunohistochemical detection of NTRK1 and NTRK2 could be useful in separating squamous cell carcinomas and adenocarcinomas in situations where histology is not definitive, particularly where the amount of tissue is restricted, such as in core biopsies. The tissue core format of the TMAs used in this study mimics the conditions of clinical core biopsy specimens and indicates that the NTRK1 and NTRK2 immunohistochemistry methodology presented here is applicable in these situations. In this study we did not directly assess the applicability of NTRK1 and NTRK2 expression in differentiating lung squamous cell carcinoma from secondary (ie, metastatic) disease; however, NTRK1 and NTRK2 staining is absent in most carcinomas present in the control tissue TMA, including carcinomas of the colon, kidney, and ovary (data not shown). These anecdotal findings suggest that NTRK1 and NTRK2 may be useful in separating primary squamous lung tumors from metastases in situations where histology is not definitive, but additional studies to properly address this possibility are required.

Other immunohistochemical markers associated with a particular line of differentiation have been proven clinically valuable, although none are both highly sensitive and highly specific. When compared with adenocarcinomas, p63 and CK5/6 are highly sensitive for squamous cell carcinomas (more than 90%), but less specific.16,18,19 CK7 is a sensitive marker for adenocarcinomas when compared with squamous cell carcinomas (more than 90%) but is less specific,18,20 whereas thyroid transcription factor 1 (TTF-1) is highly specific for adenocarcinomas (more than 95%) but typically only moderately sensitive (less than 80%).21 In this context, both NTRK1 and NTRK 2 exhibit comparatively high specificity but low sensitivity for squamous cell carcinoma, suggesting that they may be most effective in separating squamous cell carcinoma from adenocarcinoma as part of an immunohistochemical panel that includes other, more sensitive markers. We are presently exploring the utility of NTRK1 and NTRK2 in this manner by comparison with other markers including p63, CK5/6, CK7, and TTF-1.

We found the immunohistochemical expression of NTRK2 to be a significant predictor of improved DSS and OS in squamous cell lung carcinoma. NTRK2 expression as a positive prognostic indicator is independent of histologic grade, age, sex, and smoking history in the subset of squamous carcinomas for which this information is available (approximately 66% of the total squamous carcinomas). The carcinomas included in the TMAs used in this study were obtained between 1978 and 2002, and unfortunately accurate staging information could be obtained for only the most recent cases. This precluded staging from the multivariable analysis, and NTRK2 as an independently significant prognostic variable cannot be confirmed in this study; however, we do demonstrate that NTRK2 expression is positively correlated with higher stage disease, which is consistent with a recent study of NTRK2 expression in non–small cell lung carcinoma and indicates that the relationship between NTRK2 and better survival is not the result of selection for lower stage disease.10 Ultimately, prospective analysis is required to validate NTRK2 as a clinically useful positive prognostic biomarker in squamous cell carcinoma.

There are few previous studies of the immunohistochemical expression of NTRK1 and NTRK2 in lung cancer with which to compare our results. These include a recent study10 showing universal expression of NTRK2 in non–small cell lung carcinoma, but unfortunately this study did not discriminate between histologic subtypes. Another study12 found that 25% of well-differentiated squamous cell carcinomas expressed cytoplasmic NTRK2 whereas none expressed NTRK1. NTRK1 and NTRK2 staining has also been reported in well-differentiated adenocarcinomas, small cell carcinomas, bronchioloalveolar carcinomas, and the majority of carcinoid tumors.4,12–14 Some of these studies report apparently contradictory NTRK1 and NTRK2 expression profiles in normal and malignant tissues, which likely arise from preanalytic and analytic differences between these studies. Our results also differ from some of those reported in these previous studies, mostly likely for similar reasons. Importantly, our study uses standardized, commercially available monoclonal antibodies in conjunction with automated immunostaining and standard protocols, and could be easily adapted for clinical use.

The association of NTRK2 with better patient outcome in squamous cell lung carcinoma suggests an important role for NTRK2 in the biology of this tumor, and it is interesting to speculate about the underlying mechanism. During early nervous system development, nascent neurons are dependent on neurotrophin-induced signaling through NTRK receptors for survival, and limiting concentrations of neurotrophins ensure that an appropriate amount of neurons survive in a given situation and location.3 The improved outcome for patients with squamous cell carcinomas that express NTRK2 may arise from a similar function of NTRK2 in these squamous cell carcinomas, whereby they are dependent on NTRK2 signaling for survival and limiting levels of agonists limit tumor growth. This model has 2 important implications: that poor outcome in squamous carcinoma is related to circumvention of the requirement for NTRK signaling, and that inhibition of this signaling in the subset of squamous lung carcinomas defined by positive NTRK2 immunostaining may be an effective systemic therapy. Additional investigation at the molecular level is required to determine if this model is true.