Context.—Differentiation of non–small cell carcinoma into histologic types is important because of new, successful therapies that target lung adenocarcinoma (ACA). TTF-1 is a favored marker for lung ACA but has limited sensitivity and specificity. Napsin A (Nap-A) is a functional aspartic proteinase that may be an alternative marker for primary lung ACA.
Objectives.—To compare Nap-A versus TTF-1 in the typing of primary lung carcinoma and the differentiation of primary lung ACA from carcinomas of other sites.
Design.—Immunohistochemistry for Nap-A and TTF-1 was performed on tissue microarrays of 1674 cases of carcinoma: 303 primary lung ACAs (18.1%), 200 primary squamous cell lung carcinomas (11.9%), 52 primary small cell carcinomas of the lung (3.1%), and carcinomas of the kidney (n = 320; 19.1%), thyroid (n = 96; 5.7%), biliary (n = 89; 5.3%), bladder (n = 47; 2.8%), breast (n = 93; 5.6%), colon (n = 95; 5.7%), liver (n = 96; 5.7%), ovaries (n = 45; 2.7%), pancreas (n = 48; 2.9%), prostate (n = 49; 2.9%), stomach (n = 93; 5.6%), and uterus (n = 48; 2.9%). Cases were evaluated against a negative control as negative, weak positive, and strong positive.
Results.—Nap-A was more sensitive than TTF-1 for primary lung ACA (87% versus 64%; P < .001). Nap-A was more specific than TTF-1 for primary lung ACA versus all tumors, excluding kidney, independent of tumor type (P < .001).
Conclusions.—Nap-A is superior to TTF-1 in distinguishing primary lung ACA from other carcinomas (except kidney), particularly primary lung small cell carcinoma, and primary thyroid carcinoma. A combination of Nap-A and TTF-1 is useful in the distinction of primary lung ACA (Nap-A+, TTF-1+) from primary lung squamous cell carcinoma (Nap-A−, TTF-1−) and primary lung small cell carcinoma (Nap-A−, TTF-1+).
Lung cancer is the most frequently diagnosed cancer and the leading cause of cancer mortality in the world.1,2 Most lung cancers are in advanced stages when first detected, with a 5-year survival rate of 14%.2,3 Primary pulmonary tumors consist of several histologic types, most of which can be classified as malignant epithelial tumors. Of these malignant epithelial tumors, most are commonly classified as non–small cell lung carcinomas (NSCLCs) and as small cell carcinomas (SCCs). Non–small cell lung carcinoma accounts for approximately 80% of all lung cancers, with SCC accounting for most of the rest.2,4 Traditionally, the most common NSCLC consists of adenocarcinoma (≥40%) and squamous cell carcinoma (30%).1,2,4 Previously, it was sufficient to diagnosis primary lung carcinoma as either NSCLC or SCC for treatment purposes. With the development of new, successful treatments for adenocarcinoma, it is essential to diagnose the type of NSCLC whenever possible. These new treatments include (1) therapies that target EGFR mutations and ALK fusion genes that are found almost exclusively in adenocarcinomas; (2) bevacizumab (monoclonal antibody to VEGF), which is effective as a first-line agent in many adenocarcinomas but may cause severe, even life-threatening, hemorrhage in patients with squamous cell carcinoma; and (3) pemetrexed (new antifolate agent) may be effective in adenocarcinomas but is not effective in squamous cell carcinomas.5–9 The lung is also a common site of metastases from other sites, including breast, gastrointestinal tract, and genitourinary tract. Metastatic adenocarcinoma of an unknown primary accounts for approximately 3% to 5% of all malignant neoplasms and, as such, is one of the 10 most-frequently diagnosed cancers in humans.10,11
Routine sections stained with hematoxylin-eosin remain the most-common method by which lung cancers are classified; however, typing of NSCLC and the more poorly differentiated tumors is often hard to achieve by hematoxylin-eosin alone.1,12,13 Immunohistochemistry has emerged as a powerful, adjunctive tool for the differential diagnosis of lung carcinomas, whether primary or secondary to the lung. The most-useful application is in distinguishing primary lung tumors from metastatic tumors to the lung from common sites (colon, breast, prostate, pancreas, stomach, kidney, bladder, ovaries, and uterus). Although there is currently no “lung-specific tumor marker,” continued efforts to identify such a marker are scattered throughout the literature. Although progenitor stem-cell markers are being hypothesized as an immunohistochemical method to differentiate pulmonary carcinomas, that has not been well established.14 With the help of a relatively restricted marker, thyroid transcription factor 1 (TTF-1), it is possible to separate a lung primary from a metastasis with a reasonable degree of certainty. Another lung-specific marker on the horizon is napsin A (Nap-A), which appears to complement TTF-1 in defining a primary lung carcinoma,15 also appears to be helpful in subtyping NSCLC, and may be helpful in distinguishing NSCLC, particularly poorly differentiated adenocarcinoma, from SCC.1
TTF-1 is identified as a nuclear tissue-specific, 38-kDa, homeodomain protein containing DNA-binding activity. TTF-1 regulates gene expression in the thyroid, lungs, and diencephalon during embryogenesis. It is expressed in follicular cells of the thyroid, in areas of the developing brain, in type II pneumocytes, and in nonciliated bronchiolar epithelial cells.16 In the lung, TTF-1 is responsible for transcriptional activation of surfactant proteins A, B, and C and Clara cell secretory proteins. Neoplasms of pulmonary origin have been found to retain variable TTF-1 expression according to histologic type.1,15,17 Antibodies against TTF-1 antigen are well established for the differentiation between primary lung adenocarcinomas and adenocarcinomas from other sites; however, sensitivities as low as 54% in primary lung adenocarcinoma have been reported.18 TTF-1 results are also positive in 7% to 10% of primary squamous cell carcinomas of the lung.19–25 In addition, TTF-1 positivity is not universally specific to the lung. TTF-1 positivity is seen in up to 100% of primary thyroid carcinomas and in up to 90% of small cell carcinomas of any origin.19–27
Napsin A is a functional aspartic proteinase, expressed in the cytoplasm of healthy lung parenchyma. It is homologous with the polypeptide TAO2 and involved in maturation of the biologically active surfactant protein B. It also consists of a 38-kDa protein, a single-chain protein expressed in type II pneumocytes, alveolar macrophages, renal tubules, and exocrine glands and ducts in the pancreas.18,28,29 It is involved in the N-terminal and C-terminal maturation of prosurfactant protein B in type II pneumocytes.30 Earlier, we demonstrated that Nap-A is expressed in primary lung carcinomas and facilitates the differentiation between primary lung carcinomas and metastatic lesions.15 Positive immunoreactivity is seen as usually intense, granular cytoplasmic staining.1,31 Other studies have shown that Nap-A is more sensitive than TTF-1 in distinguishing lung primary from other adenocarcinomas.18,31,32
In this study, we compared Nap-A with TTF-1 in an effort to distinguish primary lung carcinoma from tumors of other organ sites, to better subtype NSCLC, and to help distinguish primary NSCLC from primary lung SCC. We used tissue microarrays to examine the staining patterns of 1674 primary carcinomas with Nap-A and TTF-1. The carcinoma types included bile duct, bladder, breast, colon, kidney, liver, lung (adenocarcinoma, squamous cell carcinoma, and small cell carcinoma), ovaries, pancreas, prostate, stomach, thyroid, and uterus. To our knowledge, our study is the largest study to date examining the sensitivity of Nap-A, compared with TTF-1, in primary lung adenocarcinoma and the largest study to date examining the expression of Nap-A in primary lung SCC.
MATERIALS AND METHODS
A total of 1674 tumor sections were processed in triplicate. Tissues studied were obtained from 2 sources. The first group of tissues studied was made available by the Laboratory of Pathology at Toyama University Hospital (Toyama, Japan). Paraffin sections (6 µm) from primary carcinomas of the bile ducts (n = 89; 5.3%), bladder (n = 47; 2.8%), breast (n = 93; 5.6%), colon (n = 95; 5.7%), kidney (n = 93; 5.6%), liver (n = 96; 5.7%), lung (n = 188; 11.2%; squamous cell [n = 94; 5.6%] and adenocarcinoma [n = 94; 5.6%]), ovaries (n = 45; 2.7%), pancreas (n = 48; 2.9%), prostate (n = 49; 2.9%), stomach (n = 93; 5.6%), thyroid (n = 96; 5.7%), and uterus (n = 48; 2.9%) were selected. A second group of tissues to be studied were made available from Weill College of Medicine Cornell University, The Methodist Hospital (Houston, Texas). These included primary lung tumors (n = 338; 20.2%; adenocarcinoma [n = 209; 12.5%], squamous cell carcinoma [n = 106; 6.3%], and small cell carcinoma [n = 23; 1.4%]), a separate set of small cell carcinomas of the lung (n = 29; 1.7%), and renal tumors (n = 227; 13.6%). The samples studied were originally fixed in 10% formalin for periods ranging from 1 to 7 days. Tissue microarrays in triplicate were constructed according to standard protocols33 using a dedicated tissue microarray instrument (Beecher TMA, Beecher instruments, Sun Prairie, Wisconsin).
Immunohistochemistry was performed at the immunohistochemistry laboratory at the University of Texas Health Science Center (San Antonio, Texas). Antibodies used included anti-human Nap-A mouse immunoglobulin G (IgG) monoclonal antibody (TMU-Ad02, 1∶100; Immuno-Biological Laboratories Co, Ltd, Takasaki, Japan) and anti-rat TTF-1 mouse IgG monoclonal antibody, known to react with human TTF-1 (8G7G3/1,1 1∶80; DakoCytomation Inc, Carpinteria, California). Protocols, reagents for antigen retrieval (Reveal Decloaker 10×), and the detection system were from Biocare Medical (Concord, California). Slides were stained with a Ventana Immunostainer (Ventana Medical Systems, Tucson, Arizona). Both Nap-A and TTF-1 immunostaining were performed on the first group of tissue specimens (from Toyama University Hospital). Only Nap-A immunostaining was performed on the second group of tissue specimens (from Weill College of Medicine). After staining, the tissue microarrays were scanned with a Spectrum/Spectrum Plus (Aperio Technologies, Inc; Bristol, United Kingdom) to a dedicated server and were analyzed by 3 pathologists independently. Staining intensity was evaluated as follows: negative, no staining to minimal light-brown to dust; weak positive, minimal, patchy, or diffuse cytoplasmic staining for Nap-A or nuclear staining for TTF-1; and strong positive, moderate to intense-brown, granular, cytoplasmic staining for Nap-A or nuclear staining for TTF-1. All results were evaluated relative to a negative control of the same tumor.
The χ2 statistic and the Fisher exact (FE) test were used to investigate whether distributions of categoric variables differ from one another. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (PLR), and negative likelihood ratio (NLR) were calculated in the usual format. If the ratio of the number of patients in disease group 1 and the number of patients in disease group 2 were not equivalent, PLR [PLR = sensitivity/(1 − specificity)] and NLR [NLR = (1 − sensitivity)/specificity] may be a better indicator of the proportion of patients with positive or negative test results who are correctly diagnosed because likelihood ratios do not depend on prevalence. Likelihood ratios predict the probability of disease, given a positive (PLR) or negative (NLR) test result.34 Data in the Tables are presented as the number of positive staining tumors versus the number of negative staining tumors in the same tumor type for Nap-A versus TTF-1 or as the number of positive staining tumors versus the number of negative staining tumors in different tumor types for Nap-A or TTF-1 alone.
The results of immunohistochemistry for Nap-A and TTF-1 are summarized in Table 1. Tables 2 through 6 show the sensitivity, specificity, PPV, NPV, PLR, and NLR of Nap-A and TTF-1 in primary lung adenocarcinoma relative to various tumor types. Positive immunoreactivity (sensitivity) was seen in 264 of 303 pulmonary adenocarcinomas (87.1%) for Nap-A, compared with 60 of 94 pulmonary adenocarcinomas (64%) for TTF-1 (P < .001 χ2 test; Tables 1 and 3); 24 of 84 primary pulmonary adenocarcinomas (28.6%) detected by Nap-A were missed on TTF-1 staining (Table 2). Excluding renal carcinomas, comparison of the number of cases staining positive with Nap-A in pulmonary adenocarcinoma versus all other tumors shows a significant difference (P < .001 χ2 and FE tests), independent of tumor type (Tables 1 and 3). All of the nonpulmonary adenocarcinomas displayed a weak-positive staining pattern (relative to the negative control), compared with a strong-positive staining in the pulmonary adenocarcinomas (Figure 1, A through D). Table 3 also shows that the specificity of Nap-A for primary lung carcinoma was 97% (compared with 90% for TTF-1), NPV was 96% (equal to TTF-1, although with a lower NLR), and PPV (PLR) was 89% (27.75), compared with 41% (6.71) for TTF-1. The likelihood ratios may be better comparative indicators in these results than are the predictive values (see “Statistical Methods”). Both markers showed decreased positive immunostaining with more poorly differentiated adenocarcinomas, consistent with previous studies.1 Positive immunoreactivity for Nap-A was only seen in 7 of 96 thyroid carcinomas (7%) for Nap-A, compared with 77 of 96 thyroid carcinomas (80%) for TTF-1 (P < .001 χ2 test; Tables 1 and 4). All the positive thyroid tumors were of the papillary type, and all displayed a weak-positive staining pattern (Figure 2, A) relative to the negative control (Figure 2, B). Comparison of the number of cases staining with Nap-A in pulmonary adenocarcinoma versus thyroid carcinoma shows a significant difference (P < .001 χ2 test). When comparing only primary lung adenocarcinoma versus thyroid carcinoma (Table 4), the specificity for lung carcinoma was 93%, compared with 20% for TTF-1; NPV (NLR) was 70% (0.14), compared with 36% (1.83) for TTF-1; and PPV (PLR) was 97% (11.95), compared with 44% (0.80) for TTF-1. Again, the likelihood ratios may be better comparative indicators here than are the predictive values (see “Statistical Methods”). No immunoreactivity was seen for Nap-A in our sample of small cell carcinomas (P < .001 FE test; Table 5). Positive immunoreactivity was seen in only 5 of 200 primary pulmonary squamous cell carcinomas (2.5%) for Nap-A and in only 2 of 94 primary pulmonary squamous cell carcinomas (2%) for TTF-1 (P > .99 FE test; Table 6). All positive pulmonary squamous cell carcinomas showed weaker staining than the positive staining of pulmonary adenocarcinomas (Figure 3, A and B). Comparison of the number of cases (Table 1) staining positive with Nap-A in pulmonary adenocarcinoma versus renal carcinoma (57%) shows significant differences in papillary, conventional, and urothelial types (P < .001 χ2 test) but not in the chromophobe type (P = .34 FE test) renal cell carcinomas. Again, all of the positive renal tumors displayed a weak-positive staining pattern (Figure 4, A) relative to the negative control (Figure 4, B). When comparing only primary lung adenocarcinoma versus renal cell carcinoma, specificity for lung carcinoma was 43%, NPV (NLR) was 78% (0.30), and PPV (PLR) was 59% (1.54). Inclusion of the renal carcinomas when comparing the number of cases staining positive with Nap-A in pulmonary adenocarcinoma versus all other tumors decreases the specificity and PPV (PLR) of primary lung adenocarcinoma to 84% and 55% (5.58), respectively, with the NPV (NLR) essentially unchanged (specificity, PPV [PLR], and NPV [NLR] of TTF-1 are also essentially unchanged).
Previous investigations of Nap-A have found that the sensitivity and specificity is at least equal to, and often greater than, that of TTF-1 in the identification of lung adenocarcinoma.1,2,15,18,32 Most of these studies have had a relatively small number of cases (n = <100) and did not address the specificity, NPV, NLR, PPV, or PLR of Nap-A relative to other tumor types. None of the previous studies adequately addressed the issue of sensitivity, specificity, predictive values, or likelihood ratios when considering the subtyping of primary lung adenocarcinoma, particularly primary lung small cell carcinoma.
To our knowledge, we have the largest study to date demonstrating that Nap-A has a higher sensitivity, specificity, NPV, NLR, PPV, and PLR for primary pulmonary adenocarcinoma than TTF-1 has (Table 3); 24 of 84 primary pulmonary adenocarcinomas (28.6%) detected by Nap-A were missed with TTF-1 staining (Table 2). Furthermore, Nap-A is useful in distinguishing primary lung carcinoma from thyroid carcinoma, unlike TTF-1 (Table 4). In addition, Nap-A distinguished primary lung adenocarcinoma from primary small cell carcinoma with 100% specificity in our study (Table 5). The lack of Nap-A in pulmonary neuroendocrine tumors has been previously noted by Bishop et al,1 but the number of cases in that work was too small to reach meaningful conclusions about its significance.1 Our study confirms that Nap-A is a specific marker when considering a differential diagnosis between primary lung adenocarcinoma and small cell carcinoma. Although Nap-A and TTF-1 show similar specificity for primary lung adenocarcinoma when compared with primary lung squamous cell carcinoma, Nap-A may be better at predicting the presence or absence of disease with a positive or negative test, respectively (Table 6). Table 7 details the immunohistochemical profiles in our study of primary lung adenocarcinoma, primary lung squamous cell carcinoma, and primary lung small cell carcinoma. A combination of Nap-A and TTF-1 would be expected to be useful in the distinction of these 3 tumors (Table 8). Primary lung adenocarcinoma will likely exhibit Nap-A positivity (Nap-A+) and will generally be TTF-1 positive (TTF-1+), but may be TTF-1 negative (TTF-1−). Primary lung squamous cell carcinoma will likely be negative for Nap-A (Nap-A−) and TTF-1−. Primary lung small cell carcinoma will likely be Nap-A− and TTF-1+. Similar to Bishop et al,1 we found that the Nap-A positivity decreased with more poorly differentiated lung tumors.
Two caveats deserve mention here, with an additional “caveat on the caveat.” First, the specificity of Nap-A for primary lung carcinoma appears to be substantially challenged only when the differential diagnosis involves renal cell carcinoma, which is a relevant concern, given that a primary metastatic site for renal cell carcinoma is the lung.35 Although there is a significant difference between the number of lung and renal tumors staining positive for Nap-A, the specificity, PPV, and PLR of Nap-A for primary lung adenocarcinoma, when examined against renal tumors only, is lower than it is when examined against other tumors. Although the clinical and radiographic history and the morphology should be adequate to distinguish the two, often the pathologist does not have that information. We agree with Bishop et al1 that, when considering renal cell carcinoma in the differential diagnosis, a panel of additional immunohistochemical markers should be used. Initially, TTF-1 would be helpful to better distinguish primary lung from primary renal, and then, a subset of additional markers may be helpful in further subclassifying the renal tumor type. Secondly, although Nap-A is clearly more sensitive and specific than is TTF-1, when examined against thyroid carcinoma, the rare thyroid carcinoma that does stain positive for Nap-A appears consistently to be the papillary type. This has been previously reported,1,2 but, at this point, the numbers are too small to come to any meaningful conclusion. Again, we agree with Bishop and collegues1 that, in the setting of a tumor positive for Nap-A with papillary architecture and “tall cell”–like features, immunohistochemistry for thyroglobulin should also be considered to exclude carcinoma of a thyroid origin. Additional studies may be helpful in clarifying this issue.
Having acknowledged the above, all of the nonpulmonary Nap-A–positive tumors stained weakly positive when compared against the negative control. In most cases the negative control actually showed minimal weak staining, making a positive result harder to interpret, and we may have actually overestimated the number of nonpulmonary tumors that showed positivity with Nap-A (Figures 1, A and B; 2, A and B; 4, A and B; and 5, A through D). The staining in the negative control may have been because of intrinsic biotin activity. In addition, the presence of macrophages and background should be considered when interpreting results. All the positive-staining pulmonary adenocarcinomas were strongly positive compared against a consistently negative (no staining) negative control. The Nap-A results were reported only after examining the tumor along with a negative control of the same tumor specimen being evaluated.
In summary, this study, the largest study of its kind to date, demonstrates that Nap-A is more sensitive and specific than TTF-1. Nap-A is better than TTF-1 in predicting the proportion of patients with positive test results (PPV, PLR) and negative test results (NPV, NLR) that are correctly diagnosed. Nap-A positivity is specific for primary lung adenocarcinoma when the major differential diagnosis includes primary small cell carcinoma. This is important because the surgical and chemotherapeutic options for these 2 tumors are markedly different. A combination of Nap-A and TTF-1 is useful in the distinction of primary lung adenocarcinoma from primary lung squamous cell carcinoma, and primary lung small cell carcinoma (primary lung adenocarcinoma, Nap-A+/TTF-1±; primary lung squamous cell carcinoma, Nap-A−/TTF1−; primary lung small cell carcinoma, Nap-A−/TTF-1+). Additional studies will be helpful in determining whether this is also true for less-poorly differentiated primary lung neuroendocrine tumors and small cell carcinomas of any origin. Consistent with previous studies and similar to most, if not all, immunohistochemical markers, Nap-A expression may be seen in other nonpulmonary tumors. The staining is generally weak and may actually be falsely positive when compared against a negative control of the same tumor specimen, whereas staining for primary lung adenocarcinoma is consistently, strongly positive, when compared against a negative control of the same tumor specimen. Consistent with previous studies, the tumors that are primarily implicated as challenging for the differential diagnosis include renal cell carcinoma and, possibly, papillary thyroid carcinoma.1,2 If either of these 2 tumors, or for that matter, any of the other tumors evaluated in this study, are part of the differential diagnosis of primary lung adenocarcinoma, the addition of TTF-1 (negative in renal and other nonpulmonary, nonthyroid carcinomas), thyroglobulin (negative in lung adenocarcinoma, positive in papillary thyroid carcinoma), and other appropriate immunohistochemical markers, in addition to clinicoradiographic correlation, should help to distinguish the differences.
From the Department of Pathology, University of Texas Health Science Center, San Antonio, Texas (Drs Turner and Jagirdar); the Department of Pathology, Weill College of Medicine Cornell University, The Methodist Hospital, Houston, Texas (Drs Cagle and Shen); the Department of Pathology, Nacogdoches Memorial Hospital, Nacogdoches, Texas (Dr Sainz); and the Laboratory of Pathology, Toyama University Hospital, Toyama, Japan (Dr Fukuoka).
The authors have no relevant financial interest in the products or companies described in this article.