Pulmonary large cell carcinoma (LCC) includes tumors not readily diagnosed as adenocarcinoma (ADC) or squamous cell carcinoma on morphologic grounds, without regard to immunophenotype, according to the World Health Organization (WHO). This ambiguous designation may cause confusion over selection of mutation testing and directed therapies. Several groups have proposed the use of immunohistochemistry (IHC) to recategorize LCC as ADC or squamous cell carcinoma; however, it remains unclear if strictly defined LCCs are a clinicopathologically distinct lung tumor subset.
—To compare the pathologic, molecular, and clinical features of 2 morphologically similar tumors: solid-subtype ADC and LCC.
Tumors were included on the basis of solid growth pattern; tumors with squamous or neuroendocrine differentiation were excluded. Solid ADC (n = 42) and LCC (n = 57) were diagnosed by using WHO criteria (5 intracellular mucin droplets in ≥2 high-power fields for solid ADC) and tested for KRAS, EGFR, and ALK alterations.
—Both solid ADC and LCC groups were dominated by tumors with “undifferentiated”-type morphology and both had a high frequency of thyroid transcription factor 1 expression. KRAS was mutated in 38% of solid ADCs versus 43% of LCCs (P = .62). One ALK-rearranged and 1 EGFR-mutated tumor were detected in the solid ADC and LCC groups, respectively. There were no significant differences in clinical features or outcomes; the prevalence of smoking in both groups was greater than 95%.
Other than a paucity of intracellular mucin, LCC lacking squamous or neuroendocrine differentiation is indistinguishable from solid-subtype ADC. We propose the reclassification of these tumors as mucin-poor solid adenocarcinomas.
Pulmonary large cell carcinoma (LCC) represents 1 of the 4 major categories of lung tumors, along with adenocarcinoma (ADC), squamous cell carcinoma (SQC), and small cell carcinoma.1 Large cell carcinoma represents 2.9% to 9% of all lung cancers in the United States, with a decreasing incidence during the past decade.1,2 Large cell carcinoma is considered to be a diagnosis of exclusion, defined by the 2004 World Health Organization (WHO) classification of lung tumors to include those tumors that lack “the cytologic and architectural features of small cell carcinoma and glandular or squamous differentiation.” If these tumors exhibit 5 or more mucin droplets in at least 2 high-power fields (HPFs), they are placed in the solid-type ADC category, otherwise they fall into the LCC category.1 This diagnosis should only be applied to resection specimens, as biopsies may represent undifferentiated areas within an otherwise differentiated tumor.1,3 By definition, LCC is poorly differentiated; however, it lacks clinical features to distinguish it from other non–small cell lung carcinomas (NSCLCs),1 and the implications of a diagnosis of LCC in terms of patient outcomes is unclear.
Also see p. 592.
The treatment of patients diagnosed with NSCLC has changed dramatically in recent years as targeted therapies have demanded a more precise classification system. Alterations in EGFR, KRAS, ALK, BRAF, and FGFR1, among others, are associated with relatively specific clinicopathologic features and may predict benefit (or lack thereof) from targeted inhibitors.4–7 In addition, determination of ADC versus SQC has implications for patient eligibility for bevacizumab and pemetrexed, both of which are approved only for patients with nonsquamous NSCLC.8,9 The use of these therapies in LCC has not been extensively tested because of the relative rarity of the diagnosis, although current recommendations lump LCC in the ADC category for treatment and molecular testing purposes.10 Recent work supports this approach. Hou et al11 showed molecular similarities between ADC and LCC, though they did not specifically focus on poorly differentiated solid-growth tumors. Botling et al12 used publicly available lung tumor gene expression data sets to show that tumors classified as LCC by histology did not distinguish themselves molecularly from ADC.
Currently, the strict definition of LCC, as defined by the WHO, does not include the use of immunohistochemistry or other ancillary studies except for mucin histochemical stains. As a result, LCC contains a heterogeneous mix of entities, which complicates any potential characterization studies or clinical trials. In addition to gene expression–based approaches,11–17 multiple studies18–27 have shown that LCC can be reclassified as ADC, SQC, or large cell neuroendocrine carcinoma by immunohistochemistry. The widespread use of immunohistochemistry in diagnostic practice may explain the decreasing incidence of LCC. Rekhtman et al18 recently reported that 80% of LCCs, as defined by solid morphology and without regard to mucin expression, could be reclassified as ADC or SQC by immunoprofile. In addition, this group found that tumor immunoprofile had important clinicopathologic implications, with thyroid transcription factor 1 (TTF-1) and ΔNp63 null tumors demonstrating significantly worse patient outcomes.
Under the WHO guidelines, many cases of solid ADC and LCC are morphologically similar and may not be consistently segregated in clinical practice. A direct comparison of the clinicopathologic and genetic characteristics of these groups, applying strict WHO criteria, is needed to determine whether this diagnostic segregation is warranted. This study hypothesizes that solid ADC and nonsquamous, nonneuroendocrine LCC are essentially one and the same. We test this hypothesis by applying the WHO definitions of solid ADC and LCC and comparing the morphology, immunophenotype, mutational status, and clinical features of these diagnostic entities.
MATERIALS AND METHODS
This study was performed after approval by the hospital's institutional review board. The pathology archives were searched for surgical lung resection specimens diagnosed as “large cell carcinoma,” “adenocarcinoma, solid subtype,” and “poorly differentiated non–small cell lung carcinoma.” To increase our capture rate, we included in our search cases diagnosed as poorly differentiated adenocarcinoma that reported areas of solid growth and “poorly differentiated squamous cell carcinoma” that did not have immunohistochemistry performed at the time of diagnosis. Tumors with the following features were excluded from the analysis: at least 10% glandular, sarcomatoid, or giant cell differentiation; strong p63 and/or p40-positive/TTF-1–negative immunophenotype; morphologic and immunophenotypic characteristics of large cell neuroendocrine carcinoma; or preoperative chemotherapy or radiotherapy.
Staining for hematoxylin-eosin, mucicarmine, TTF-1, and p63 immunohistochemistry (IHC) was reviewed in each case included in the study. Weak to moderate p63 staining in TTF-1–negative or weak cases was permitted if subsequent p40 staining was negative or only present in rare cells. Distinction between solid ADC and LCC was made according to WHO criteria (5 intracellular mucin droplets in at least 2 HPFs for solid ADC). Tumors with focal glandular architecture (<10%) were categorized as solid ADC regardless of the presence or absence of mucin.28 Clear cell features were recorded as absent (no clear cell features), focal (<50% clear cell features), or extensive (50% or greater clear cell features). Patient demographics and clinical outcomes were derived from the electronic medical record and the Social Security Death Index.
The primary antibodies used were TTF-1, p63, p40, and anaplastic lymphoma kinase (ALK). The clones, dilutions, and pretreatment conditions are shown in Table 1. The Dako Envision Plus System (Dako North America, Inc, Carpinteria, California) was used for signal detection. Nuclear staining was considered positive for TTF-1, p63, and p40. Cytoplasmic staining was considered positive for ALK.
Tumor was macrodissected from unstained 10-μm sections or was cored from the corresponding paraffin-embedded blocks by using a 1-mm punch. DNA was extracted by using the QIAGEN QIAamp DNA Mini Kit (51304, QIAGEN, Germantown, Maryland).
All cases were tested for KRAS codon 12 and 13 mutations. Following amplification by polymerase chain reaction (PCR), pyrosequencing (QIAGEN) was performed with 2 different sequencing primers to detect all possible nucleotide substitutions in codons 12 and 13 (primer sequences available on request).
Tumors wild type for KRAS were subsequently tested for EGFR mutations. The EGFR c.2573T>G(p.Leu858Arg) point mutation was tested by using allele-specific amplification with a TaqMan-based real-time PCR detection system (ABI, Life Technologies, Grand Island, New York). EGFR exon 19 deletion mutations were detected by using a PCR-based sizing assay (primer sequences available upon request).
Clinicopathologic and pathologic characteristics were compared by using a χ2 test and when appropriate, Student t test and 2-tailed Fisher exact test. Progression-free and overall survivals were calculated by using a Kaplan-Meier curve and the group comparisons were performed by using a log-rank test, taking censoring into account. Progression-free and overall survival times were calculated by subtracting date of diagnosis from date of disease recurrence or death, respectively. Multivariate survival analysis was performed by using Cox proportional hazards regression (Stata version 12.1, College Station, Texas).
A search of the archives at our hospital identified 285 candidate cases from 1989 to 2012. After excluding those cases with neoadjuvant treatment, insufficient/unavailable material for morphologic review and ancillary testing, and those that proved to be of nonlung origin, 134 cases remained for further analysis. After morphologic and immunohistochemistry review, 7 were classified as SQC, 8 as large cell neuroendocrine carcinoma, 13 as mixed subtype ADC, and 7 as sarcomatoid carcinoma; these cases were excluded from further analysis.
The remaining 99 cases included 42 cases of solid ADC and 57 cases of LCC. The clinical characteristics of these 2 groups are shown in Table 2. Upon review of histology and TTF-1, p63 +/- p40, and mucicarmine stains for this study, 10% of tumors originally reported as LCC were changed to solid ADC and 28% of cases originally diagnosed as solid ADC were changed to LCC. Greater than 95% of patients in both groups were ever smokers. Most cases in both groups represented stage I or II tumors. There were no statistically significant clinical differences between the 2 groups.
Morphologic and Immunohistochemical Characteristics
The morphologic and immunohistochemical characteristics for these cases are shown in Table 3 and Figure 1. Most cases in both solid ADC (Figure 1, A and B) and LCC (Figure 1, C and D) groups had undifferentiated morphology,1 comprised of large, polygonal cells with moderate eosinophilic to clear cytoplasm. Some degree of clear cell change was common and did not correlate with diagnosis (Figure 2, A). A minority of undifferentiated tumors showed focal to extensive nested growth with peripheral palisading (Figure 2, B and C), a feature that was more common in the solid ADC category (21% versus 2%, P = .002). Basaloid growth pattern was rare and in most cases did not have associated mucin production (Figure 2, D). While this was most commonly a pure pattern, in a few cases basaloid growth was seen together with undifferentiated (Figure 3, A and B) and palisading growth.
In both groups, most cases were TTF-1 positive with absent to variable p63 expression (69% versus 82% in solid ADC versus LCC, P = .12). Two solid ADCs and 1 LCC with absent TTF-1 and focal, weak p63 staining were p40 negative. Of the basaloid tumors, 3 were TTF-1 positive (Figure 3, C), 3 were TTF-1 negative, and all were p63 negative (Figure 3, D). TTF-1 expression was seen in nearly all tumors with palisading features (11 of 12, 92%), but this was not significantly different from the frequency of TTF-1 in tumors without palisading (65 of 87, 75%, P = .29).
Mutation status could be determined in 95 cases; 4 cases failed mutational analysis. KRAS mutations were present in 38% and 43% of solid ADC and LCC groups, respectively, most of which were transversion mutations (Table 4). Overall, KRAS mutations were detected at a similar frequency in tumors with palisading (2 of 9, 22%), any basaloid growth (2 of 5, 40%), and pure undifferentiated growth (35 of 81, 43%) (not significant). There was 1 ALK-rearranged tumor in the solid ADC group and 1 EGFR p.L858R-mutated tumor in the LCC group (Figures 4 and 5).
The single ALK-rearranged tumor was small (stage IA) and came from a 50-year-old woman with a 20-pack-year smoking history. The patient was treated with lobectomy alone and was lost to follow-up after 7 months. On review, the tumor exhibited solid growth with focal cribriform pattern and abundant intracellular and extracellular mucin (Figure 4, A) and strong ALK protein expression by immunohistochemistry (Figure 4, B).
The single patient with an EGFR-mutated tumor was a 62-year-old woman with a 100+-pack-year smoking history who presented with a solitary brain metastasis and underwent metastatectomy and lung lobectomy. This patient received adjuvant radiation and chemotherapy, and was doing well at 5 months of follow-up. The tumor had a histologically undifferentiated LCC phenotype (Figure 5) and strong TTF-1 expression.
Mean progression follow-up time was 33 months (range, 1–286 months). Median progression-free survival (PFS) was not reached for either the solid ADC or LCC groups. Patients with a solid ADC diagnosis had a 71% 5-year PFS, as compared to 58% in the LCC group. The Kaplan-Meier survival curves separate at the 5-year time point for the solid ADC and LCC groups (Figure 6, A); however, this difference is not statistically significant (log-rank test, P = .21). On multivariate analysis only stage grouping was a significant predictor of PFS (Table 5). The overall survival (OS) was nearly identical in the 2 groups (54% 5-year OS with solid ADC versus 49% with LCC, log-rank test, P = .96, Figure 6, B), with a mean OS follow-up time of 45 months.
Correlation Between TTF-1 Expression and Other Clinicopathologic and Genetic Features
In light of prior studies of LCC that have reported a significant association between absent TTF-1 expression and poor patient outcome, we also correlated clinicopathologic variables and clinical outcomes with TTF-1 expression in the tumors. As noted above, TTF-1 expression did not correlate with diagnosis or with histologic pattern. Other variables, including patient age, sex, smoking pack years, stage, and KRAS mutation status, failed to predict the presence of TTF-1 expression (data not shown). We then examined the effect of TTF-1 expression on outcomes in the LCC and solid ADC subgroups, and found it had no significant association with PFS or OS (log-rank test, P = .52 and .25, respectively, for PFS; and P = .42 and .78, respectively, for OS). Because our prior analysis indicated there is little difference between the solid ADC and LCC categories, we then combined these categories and examined the effect of TTF-1 expression on outcomes in this larger data set. TTF-1 expression had no impact on PFS by univariate analysis (P = .62; Figure 6, C) or multivariate analysis (Table 5), or on OS (P = .26; Figure 6, D) by univariate analysis in the overall cohort.
Clear delineation of lung carcinoma subtype is of paramount importance in guiding mutation analysis and choice of therapy. The widespread use of immunohistochemistry and our growing knowledge of tumor genetic signatures can complement the pathologist's morphologic assessment and promote more clinically useful categorization of histologically ambiguous tumors. In light of the acknowledged morphologic overlap between solid-subtype ADC and a large subset of LCCs, we hypothesized that these tumor types would share other clinical and genetic features.
In contrast to previously published studies,13,18 we applied strict WHO criteria to categorize lung carcinomas with a solid growth pattern into solid ADC and LCC. Squamous cell, pleomorphic, and large cell neuroendocrine tumors were carefully excluded, using immunohistochemical stains when necessary. We then compared the clinical, pathologic, and genetic features of solid ADC and LCC and determined that they are virtually indistinguishable.
The core definition of LCC relies on the absence of clear glandular or squamous features and the absence of any significant amount of intracellular mucin droplet formation. In reviewing our internal practice, the quantitative bar set by the WHO of greater than 5 mucin droplets in 2 HPFs for solid ADC was not uniformly applied in the time period examined, likely owing in part to the fact that diagnostic practices have changed significantly over the time period examined (1989–2012) and owing to inconsistent application of mucin histochemistry. The use of solid ADC and LCC diagnoses interchangeably in practice is not surprising in light of the high degree of morphologic overlap between these categories. As has been previously reported,18 we observed a range of morphologies, even within these subgroups of tumors with solid growth patterns. Undifferentiated tumors were most common, a subset of which showed nested growth with peripheral palisading. This was more common among solid ADC and in some cases was associated with extensive mucin production (Figure 2, C); ultimately, the presence of palisading features was the only distinctive variable between solid ADC and LCC groups. The basaloid pattern of growth, seen here in 6% of cases overall, is more commonly associated with squamous differentiation and, according to the WHO, is TTF-1 negative.1 However, 3 (50%) of the basaloid tumors described here were strongly TTF-1 positive. Interestingly, 3 cases had coexisting basaloid and undifferentiated components, suggesting that these patterns exist on a continuum.
We observed KRAS mutations in approximately 40% of LCCs and solid ADCs. This frequency is higher than the 25% reported in lung ADC from unselected patients,29 likely reflecting the high numbers of smokers in these cohorts.30–34 KRAS transversion mutations in particular have been linked to smoking history in prior studies30,35,36 ; among the KRAS mutations detected in our study, approximately 90% were transversions, confirming this association. Because KRAS mutations are mutually exclusive with the other common lung ADC alterations in EGFR and ALK, our testing strategy restricted EGFR and ALK testing to the KRAS wild-type tumors. Among the KRAS wild-type tumors, 1 ALK-rearranged solid ADC and 1 EGFR-mutated LCC were identified, for a frequency of 1% for each of these genetic alterations in the overall cohort. These observations suggest some role for clinicopathologic profiles directing the order of mutational analysis, particularly in clinical diagnostic laboratories using single gene assays in environments sensitive to time and cost. However, the presence of a significant smoking history in both of these patients confirms prior recommendations that testing for EGFR and ALK alterations, both of which are associated with never-smoking status and predict response to targeted therapies, should not be limited by clinical selection.10
Solid pattern histology is a poor prognostic indicator in lung adenocarcinomas.28,37 No statistically significant difference in PFS was noted between the solid ADC and LCC groups. However, because of the bias toward low stage in this cohort of surgically resected tumors, and because of censoring from death from other causes (as reflected in the overall survival data) or because patients were lost to follow-up, the number of disease events in both groups was relatively low, with neither group reaching a median PFS point. The nearly identical OS in the 2 groups may be a manifestation of similar comorbidities related to significant smoking histories, although this kind of clinical data were not readily available for many of the patients in this retrospective study. Keeping these caveats in mind, this study suggests there is no difference in outcomes between mucin-producing solid ADC and mucin-poor “LCC.”
Multiple reports have shown that TTF-1 expression is a good prognostic factor in lung ADCs, most of which include tumors with a mix of histologic patterns and do not focus on poorly differentiated solid tumors.38–42 Solis et al41 noted that while they were able to show a relationship between TTF-1 expression and better prognosis in nonsolid tumors, this relationship did not hold true for tumors that had a component of solid growth. Rekhtman et al,18 examining a total of 102 LCCs, 20 of which were TTF-1 negative, have shown that the absence of TTF-1 or ΔNp63 in LCC (immunomarker “null”) is associated with a distinctly worse disease-free and overall survival, as compared to LCC with TTF-1 or ΔNp63 expression. We examined the effect of TTF-1 expression on survival in the LCC and solid ADC groups independently, and then, to increase our power, examined it in the entire cohort; however, we were unable to reproduce these observations in our study. This discrepancy likely reflects the limited power of both studies; further studies with larger numbers of TTF-1 negative “large cell carcinomas” and/or meta-analyses will be needed to resolve this discrepancy.
In conclusion, our study suggests that nonsquamous, nonneuroendocrine LCCs with undifferentiated morphology can justifiably be recategorized as mucin-poor solid adenocarcinomas. Immunohistochemical studies are warranted to separate these poorly differentiated tumors from SQCs; however, further studies are needed to determine if LCC with squamous immunophenotype is clinicopathologically distinct from morphologically diagnostic squamous carcinoma. The absence of TTF-1 expression should prompt some consideration of spread from other primary sites, keeping in mind that the sensitivity of TTF-1 expression for lung adenocarcinomas is only 70% to 80%.43 Further studies are needed to determine if the basaloid tumors described here are a distinct clinicopathologic entity or are in fact on a genetic and pathologic continuum with other solid-subtype adenocarcinomas, as our data would suggest. As more advanced molecular tools become commonplace, identification of novel genetic alterations will undoubtedly aid in these classification efforts and should be harnessed to further refine our pathologic diagnoses.
The authors have no relevant financial interest in the products or companies described in this article.