Large cell neuroendocrine carcinoma of the lung

INTRODUCTION — Large cell neuroendocrine carcinoma (LCNEC) is a rare pulmonary tumor, with features of both small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC).

Due to the rarity of LCNEC, there are no large randomized trials that define the optimal treatment approach for either localized or advanced disease [1]. Treatment recommendations are based upon extrapolation from the approach to patients with SCLC, NSCLC, and the established literature, which is primarily retrospective in nature. Our suggested approach to LCNEC is discussed here.

Discussions dealing with SCLC, NSCLC, and the pathology of lung cancer are found elsewhere.

●(See "Extensive-stage small cell lung cancer: Initial management".)

●(See "Limited-stage small cell lung cancer: Initial management".)

●(See "Treatment of refractory and relapsed small cell lung cancer".)

●(See "Overview of the initial treatment of advanced non-small cell lung cancer".)

●(See "Initial management of advanced non-small cell lung cancer lacking a driver mutation".)

●(See "Pathology of lung malignancies".)

CLINICAL PRESENTATION — The clinical presentation of LCNEC appears similar to the other high-grade neuroendocrine pulmonary tumor, small cell lung cancer (SCLC), with notable exceptions: primary LCNECs tend to be located peripherally rather than centrally, and presentation of LCNECs with early-stage (I to II) disease is more common than for SCLC (approximately 25 versus less than 5 percent). Thus, patients with LCNEC more commonly undergo resection.

Further, because LCNEC can be a difficult diagnosis based on needle aspirate or small biopsy, the diagnosis is frequently made post-resection. Approximately 25 percent of LCNEC tumors are combined with other histologies, including SCLC most commonly, as well as adenocarcinoma and squamous carcinomas.

MOLECULAR FEATURES — LCNEC harbors a high mutation burden (8.5 to 10.5 mutations/Mb) and the high transversion rate associated with smoking [2,3]. It should also be noted that infrequently (5 to 7 percent), LCNEC tumors may harbor potentially actionable alterations including mutations in KRAS, EGFR, or MET and rearrangement of EML4-ALK [4]. (See 'Special considerations for those with actionable driver mutations' below.)

Is LCNEC more like SCLC or NSCLC?

Molecular profiling — Genomic profiling of LCNEC using formalin-fixed paraffin-embedded (FFPE) tissue and targeted sequencing panels suggests similarities between both small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC) with regards to specific mutations [2,5,6]. However, the first investigation using integrative genomic and transcriptomic profiling of LCNEC shows that, despite mutations in genes commonly seen in NSCLC, the overall transcriptional pattern is most similar to SCLC [7].

A genomic study using next-generation sequencing of 241 genes in 45 LCNEC resected tumors, demonstrated two major genomic signatures [2].

●NSCLC-like (RB1 WT +/- KRAS, STK11, KEAP-1 mutant) – Twenty-five (55 percent) of tumors were TP53 mutant, but RB1 wildtype (WT). Twenty-four (96 percent) of these RB1 WT tumors carried mutations commonly seen in NSCLC including Kirsten rat sarcoma viral oncogene homolog (KRAS), serine/threonine kinase 11 (STK11), and Kelch-like ECH-associated protein 1 (KEAP1).

Notably, however, this latter group also had frequent mutations in notch homolog 1 to 4 (NOTCH1-4), similar to SCLC.

●SCLC-like (TP53/RB1 altered) – Eighteen tumors possessed alterations commonly found in SCLC, including coalterations of tumor protein p53 (TP53), retinoblastoma gene 1 (RB1; 18 of 43, 42 percent), and L-myc-1 proto-oncogene (MYCL) amplification.

These two genomically defined subsets, primarily segregating by RB1 status, were also identified by others [4-6]. For example, whole-exome sequencing of 467 cases of LCNEC found that 79.1 and 36.8 percent of tumors had alterations in TP53 and RB1, respectively, the two most commonly altered genes [4]. In this cohort, 24 percent of LCNECs were considered genomically SCLC-like (both TP53 and RB1 altered), while the remaining 76 percent were NSCLC-like [4].

Subsequently, a 2018 study reported the first integrated genomic and transcriptomic profiling of LCNEC with comparison to other lung histologic subtypes [7]. Whole-exome or whole-genome sequencing of 60 tumor-normal pairs, RNASeq on 69 tumors, and single-nucleotide polymorphism (SNP) arrays for copy number variations on 60 tumors were performed. They confirm the prior reports in showing two genomic subgroups segregating based primarily on RB1 status. These groups were termed type 1 LCNEC (biallelic alterations of TP53 and STK11/KEAP1), RB1 WT (37 percent), and type II LCNEC (biallelic inactivation of TP53 and RB-1; 42 percent), similar to the NSCLC-like and SCLC-like subsets, respectively, discussed above.

However, comparison of the transcriptomic profiles of LCNEC to existing gene expression data for SCLC, pulmonary carcinoids, squamous cell carcinomas, and adenocarcinomas, showed that LCNEC clustered in transcriptional subgroups with SCLC, one cluster predominantly LCNEC, and one cluster predominantly SCLC. There was no clustering of LCNEC with either adenocarcinomas or squamous cell carcinomas. Thus, the profiles were distinctly different from the NSCLC tumors, despite harboring similar mutation signatures. They were much more similar to SCLC, though in many respects unique. Differential gene expression analysis confirmed that both type I and type II LCNECs commonly showed upregulation of neuroendocrine and endocrine markers and lineage-specific transcription factors commonly upregulated in SCLC (eg, ASCL1 and NeuroD1). There was also upregulation of E2F targeted genes, checkpoint kinases, and DNA damage response genes, confirming that LCNEC is truly a neuroendocrine tumor and not a neuroendocrine version of NSCLC subtypes. Similar to the majority of SCLCs, type I LCNECs, despite carrying mutations common to NSCLC, have an expression profile with ASCL1 high/DLL3 high/NOTCH low. Type II LCNECs, which carry the hallmark p53/RB1 null profile of the SCLC, paradoxically exhibit decreased expression of neuroendocrine markers, ASCL1 low/DLL3 low/NOTCH high, and increased expression of immune-related genes. While the transcriptomic data should be considered preliminary until confirmed in additional cohorts, they do provide a framework for further investigation, highlighting the importance of transcriptional profiling, analogous to the proposed classification for SCLC that is based on four key transcription factors [8].

In summary, despite earlier genomic data suggesting "NSCLC" and "SCLC"-like genomic profiles for LCNEC, the transcriptional profile of the NSCLC-like genomic subtype of LCNEC, appears similar to the most common subtype of SCLC. Meanwhile, the SCLC-like genomic subtype is characterized by a transcriptional profile uncommon in SCLC. Overall, these findings reinforce that LCNEC is a unique and separate entity that defies simplistic classification based only on mutation pattern, and the paradigm of categorizing it as either resembling SCLC or NSCLC is inaccurate.

DIAGNOSIS — LCNEC is diagnosed based on high-grade features (>10 mitotic figures in 2 mm² of viable tumor), and the presence of both neuroendocrine morphology as well as immunohistochemical evidence of neuroendocrine markers. However, application of the latter requirement to cytologic specimens and small biopsies is problematic, and we obtain resected tumors for analysis when LCNEC is suspected, when feasible [1,9,10].

LCNECs do not meet the criteria for the three better recognized neuroendocrine tumors (carcinoid, atypical carcinoid, and small cell carcinoma). Previously classified as a variant of large cell carcinoma, LCNEC is grouped with the pulmonary neuroendocrine carcinomas in the 2015 World Health Organization classification. (See "Pathology of lung malignancies", section on 'Large cell neuroendocrine carcinoma'.)

MOLECULAR TESTING — Similar to non-small cell lung cancer (NSCLC), we suggest sequencing using a next-generation sequencing panel to identify potentially targetable alterations. This and other methods to assess tumors for genetic alterations are similar to that used in NSCLC and are discussed elsewhere (see "Personalized, genotype-directed therapy for advanced non-small cell lung cancer", section on 'Molecular testing'). It is also reasonable to test programmed cell death ligand 1 (PD-L1) expression via immunohistochemistry, as in NSCLC, although data are unclear as to whether this can inform treatment decisions in LCNEC.

Actionable mutations — Although rarer than in NSCLC, similar actionable genomic alterations are observed in LCNEC. In a large cohort of 467 LCNEC tumors profiled by whole-exome sequencing, clinically actionable alterations were seen in >11 percent of patients. (See 'Special considerations for those with actionable driver mutations' below.)

●For example, KRAS G12C mutations (2.9 percent), MET exon 14 skipping mutations (2.4 percent), ALK fusions (1.7 percent), and EGFR exon 19 deletions or L858R mutations (approximately 1 percent) were observed [4]. Thus, it is our routine practice to perform molecular profiling on LCNEC tumors with tissue-based and/or circulating tumor DNA approaches.

A special note should be made that while the aforementioned actionable alterations may be observed in treatment-naïve LCNEC, transformation to LCNEC histology has also been proposed as a mechanism of resistance to targeted therapy in driver-mutant NSCLC [11,12].

PD-L1 and immune phenotype — Expression of programmed cell death ligand 1 (PD-L1) in LCNEC has been reported in approximately 10 to 20 percent of cases [3,13].

For example, one report has described the association of PD-L1 expression with tumor-infiltrating immune cells (including lymphocytes, dendritic cells, and monocytes) and mutation burden in 192 patients [3]. This analysis included comparative data in 120 patients with small cell lung cancer (SCLC). Tumor cell PD-L1 expression ≥1 percent was found in 17 percent of LCNEC cases, similar to the 15-percent rate in SCLC. However, immune cell infiltration was more frequent in LCNEC versus SCLC (58 versus 23 percent), as was PD-L1 expression on immune cells (46 versus 23 percent). Similar rates of PD-L1 expression were observed in a cohort of 467 patients using 22C3 pharmDx immunohistochemistry assay, with 21.5 percent of samples staining >1 percent [4].

Immune cell infiltration was more frequent in early-stage versus late-stage LCNEC (61 versus 28 percent), though stage-dependent differences were not observed in SCLC. It should be noted that only 8 (7 percent) cases of SCLC were stage I or II, compared with 29 (40 percent) cases of LCNEC and thus stage effects, as opposed to histology may confound the finding that immune cell infiltration was more frequent in LCNEC. Further, technical issues could contribute to stage-dependent immune cell infiltration due to the evaluation of resected tumors in early stage, as opposed to biopsies for stages III to IV.

In a subset of cases (39 LCNEC and 63 SCLC) with tumor evaluable for both PD-L1 by immunohistochemistry and mutation burden from targeted next-generation sequencing, the number of nonsynonymous mutations/tumor was positively correlated with immune cell infiltration and PD-L1 expression on immune cells for both LCNEC and SCLC [1,10,14].

APPROACH TO TREATMENT

Due to the rarity of LCNEC, there are no large randomized trials that define the optimal treatment approach for either localized or advanced disease [1]. Treatment recommendations are based upon extrapolation from the approach to patients with small cell lung cancer (SCLC), non-small cell lung cancer (NSCLC), and the established literature, which is primarily retrospective in nature.

Early-stage disease — For patients with early-stage LCNEC, surgical resection is indicated when feasible. (See "Management of stage I and stage II non-small cell lung cancer".)

Because of the overall poor prognosis, adjuvant chemotherapy may offer the best chance to improve survival [1,15-17] regardless of stage. Our approach is to use an SCLC regimen such as etoposide plus a platinum compound for four cycles, or, alternatively, irinotecan plus a platinum. In a trial conducted in Japan in 221 patients with completely resected stage I to IIIA high-grade neuroendocrine carcinoma of the lung, etoposide and cisplatin versus irinotecan and cisplatin resulted in similar progression-free survival (PFS) rates at three years in the overall trial, as well as in the subset of 104 patients with LCNEC [18].

A meta-analysis also of 1458 LCNEC cases treated with surgery plus adjuvant chemotherapy or surgery alone generally supports this approach [19]. Odds ratios (ORs) of mortality favored adjuvant chemotherapy (OR 0.65, p <0.01) across all stages. Additional subgroup analyses showed OR of 0.68 for all stage I tumors (p = 0.02), with further subdivision into stage IA (OR 0.88, p = 0.22) or stage IB (OR 0.49, p <0.01). This analysis also compared studies of SCLC regimens (platinum plus etoposide or irinotecan) and NSCLC regimens (platinum with gemcitabine, pemetrexed, docetaxel, paclitaxel, or vinorelbine) as adjuvant therapies, with results favoring SCLC regimens after two years (OR 0.32, p = 0.03). (See "Limited-stage small cell lung cancer: Initial management", section on 'Chemotherapy' and "Limited-stage small cell lung cancer: Initial management", section on 'Alternatives'.)

Prophylactic cranial irradiation is not used as a routine practice in these patients. Limited retrospective data suggest the incidence of brain metastases in patients with recurrence is similar to that in extensive SCLC [1]. (See "Prophylactic cranial irradiation for patients with small cell lung cancer".)

For patients with positive mediastinal nodes (stage III) after resection and for those with unresectable stage III disease, our approach is to use chemoradiotherapy with etoposide/cisplatin followed by additional chemotherapy for a total of four cycles, identical to management of limited SCLC. (See "Limited-stage small cell lung cancer: Initial management", section on 'Definition of limited-stage disease'.)

Ongoing trials are investigating the use of immunotherapy concurrent with and following chemoradiotherapy for SCLC. However, as there are no existing data to support this approach even in limited-stage SCLC, we do not routinely use immunotherapy in conjunction with chemoradiation in LCNEC, including durvalumab as consolidation after completion of chemoradiation.

Advanced disease

Components of treatment

Chemotherapy — For patients with stage IV disease, we suggest using a standard SCLC regimen (etoposide plus a either carboplatin or cisplatin) for four to six cycles, independent of retinoblastoma gene 1 (RB1) status.

However, prognosis for this tumor type is poor regardless of choice of chemotherapy, and other options also are reasonable.

Our approach is based on integrative profiling that clearly identifies LCNEC to be a neuroendocrine tumor most similar to SCLC [7]. (See "Extensive-stage small cell lung cancer: Initial management" and 'Molecular profiling' above.)

Data from a prospective phase II trial of 29 patients treated with etoposide/cisplatin along with review of published retrospective experience with chemotherapy in stage IV LCNEC demonstrated that major efficacy endpoints were similar to SCLC, with the exception that overall response rate is lower (34 percent) than the approximate 60 percent rate observed in SCLC [20].

However, retrospective analysis of 79 stage IV LCNEC cases profiled with a limited next-generation sequencing panel showed that for patients with RB1 wildtype (WT) tumors, pairing a platinum agent with either a taxane or gemcitabine was associated with improved overall survival (OS) relative to using etoposide or pemetrexed with the platinum agent [5]. No statistically significant differences were seen in OS for patients with RB1-mutant tumors when pairing different agents with platinums.

Interpretation of this analysis is limited based on the small number of patients in many of the groups, the retrospective nature of the study, and the relatively poor OS rates observed for all subgroups, except the RB1 WT group treated with gemcitabine or taxane. Additionally, the authors do not report comparative response data for the two regimen types in that group, and median PFS differed only by 0.4 months (6.1 versus 5.7). Furthermore, the median PFS of 5.7 months for etoposide-treated patients is nearly identical to the median OS of 5.8 months in that group, suggesting that variables other than the platinum partner may have influenced survival. Finally, a test for interaction of chemotherapy type and RB1 mutation status was not statistically significant. Thus, guiding treatment based on these data seems premature.

Immune checkpoint blockade — Pending further data, we suggest addition of anti-programmed cell death protein 1 (PD-1) or programmed cell death ligand 1 (PD-L1) therapy to etoposide/platinum in the front-line setting. This approach is based on the promising, albeit minimal, data with immune checkpoint blockade in LCNEC discussed below, and the existing data with immunotherapy for SCLC. In the same vein, we suggest anti-PD-1 therapy, with or without anti-cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) therapy, in the second and greater lines of treatment, if it has not already been administered in an earlier treatment line. This assumes third-party reimbursement or drug access through an industry-sponsored program. There are no data to support decisions regarding immunotherapy treatment in LCNEC to be made on the basis of PD-L1 expression by immunohistochemistry. (See "Extensive-stage small cell lung cancer: Initial management", section on 'Atezolizumab plus carboplatin and etoposide'.)

Published experience with anti-PD-1 with or without anti-CTLA-4 antibodies in LCNEC is quite limited.

●A response rate of 60 percent (6 of 10) and median PFS of 57 weeks was observed in a retrospective cohort of 10 patients with LCNEC treated with nivolumab (9 patients) or pembrolizumab (1 patient) [21]. Five of 10 patients were treated in the third line or greater. Severe immune toxicity was limited to 1 patient with pneumonitis.

●The only prospective data reported are from the neuroendocrine cohort of the DART trial (NCT02834013), an ongoing basket trial evaluating the combination of nivolumab and ipilimumab in patients with rare tumors [22]. Of 33 patients accrued, 19 had high-grade histology, 3 of whom had lung primaries consistent with LCNEC. Two of these three patients (66 percent) had confirmed partial responses, while the response rate overall for patients with high-grade tumors was 8 of 19 (42 percent). Interestingly, there were no confirmed responses in patients with low- or intermediate-grade NEC. Alanine transaminase (ALT) elevation (9 percent) was the most common grade 3 or 4 immune toxicity. Nausea and fatigue were the most common toxicities occurring in approximately 30 percent of patients. Given the established high tumor mutation burden (TMB) in LCNEC [2,7], and the previously reported correlation of high TMB and response to nivolumab with and without ipilimumab in SCLC [23], it is tempting to speculate that this could explain the high response rate observed in the DART trial. Correlative immune biomarker analysis from this cohort is awaited. The DART trial will also accrue an additional cohort of patients with high-grade NEC to confirm their initial findings.

Special considerations for those with actionable driver mutations — For patients whose tumors are identified as having KRAS, EGFR, ALK, or other actionable driver mutations, data are limited, and in this setting, treatment choices depend on line of treatment, pace of disease, as well as patient functional status and preference. Patients with heavier disease burden or a more rapid pace of disease may benefit from chemotherapy and immunotherapy, as discussed above. However, for those with limited disease burden or more indolent disease, or for those wishing to avoid the toxicities associated with these treatments, we suggest a targeted therapy approach, analogous to driver-mutated NSCLC. Case reports predominantly from Asia have demonstrated variable sensitivity of LCNEC to targeted therapy [24-27]. (See 'Components of treatment' above and "Personalized, genotype-directed therapy for advanced non-small cell lung cancer".)

PROGNOSIS — Most studies indicate that stage-matched survival is worse for LCNEC compared to non-small cell carcinoma and other large cell carcinomas, and similar to that of small cell lung cancer (SCLC) [1,10].

In a surveillance epidemiology and end results (SEER) database study of 1444 early stage patients who underwent resection only (no radiation therapy and no information regarding chemotherapy), a numerically worse four-year survival rate was observed in patients with SCLC, compared with LCNEC and other large cell carcinomas [14]. However, multivariate analysis demonstrated no significant difference in overall or lung cancer-specific survival between LCNEC and SCLC, nor between LCNEC and other large cell carcinomas. These data should not alter the general impression that LCNEC carries a relatively poor prognosis, given the limitations of retrospective analysis [1]. (See "Clinical manifestations of lung cancer" and "Overview of the initial treatment and prognosis of lung cancer".)

SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Diagnosis and management of lung cancer".)

SUMMARY AND RECOMMENDATIONS

●Large cell neuroendocrine carcinoma (LCNEC) is a rare pulmonary tumor, with features of both small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC). (See 'Introduction' above.)

●The clinical presentation of LCNEC appears similar to SCLC, with notable exceptions: primary LCNECs tend to be located peripherally rather than centrally, and presentation of LCNECs with early-stage (I to II) disease is more common than for SCLC. (See 'Clinical presentation' above.)

●Although genomic profiling of LCNEC suggests similarities between both SCLC and NSCLC with regards to specific mutations, subsequent integrative genomic and transcriptomic profiling of LCNEC shows that the overall transcriptional pattern is most similar to SCLC. (See 'Molecular profiling' above.)

●LCNEC is diagnosed based on high-grade features (>10 mitotic figures in 2 mm² of viable tumor) and the presence of both neuroendocrine morphology as well as immunohistochemical evidence of neuroendocrine markers. (See 'Diagnosis' above.)

●For patients with early-stage LCNEC, surgical resection is indicated when feasible. Because of the overall poor prognosis, we suggest adjuvant chemotherapy rather than observation (Grade 2C), extrapolating from the SCLC setting. Our approach is to use a SCLC regimen such as etoposide plus a platinum compound for four cycles. (See 'Early-stage disease' above.)

●For patients with positive mediastinal nodes (stage III) after resection, and for those with unresectable stage III disease, we suggest chemoradiotherapy with etoposide/cisplatin followed by additional chemotherapy for a total of four cycles, rather than chemotherapy alone (Grade 2C). This mimics management of limited SCLC. (See 'Early-stage disease' above.)

●For patients with stage IV disease, our approach is as follows:

•We typically suggest using a standard SCLC regimen (ie, etoposide plus either carboplatin or cisplatin) for four to six cycles rather than an NSCLC regimen (Grade 2C), recognizing that data in this area are extremely limited, and that NSCLC regimens may also be considered reasonable alternatives. (See 'Chemotherapy' above.)

•We also suggest addition of anti-programmed cell death protein 1 (PD-1) or programmed cell death ligand 1 (PD-L1) therapy to etoposide/platinum in the front-line setting (Grade 2C). This approach is based on the promising, albeit minimal, data with immune checkpoint blockade in LCNEC discussed above, and the existing data with immunotherapy for SCLC. (See 'Immune checkpoint blockade' above.)

•For patients whose tumors are identified as having actionable driver mutations, data are limited, and in this setting, treatment choices depend on line of treatment, pace of disease, as well as patient functional status and preference. (See 'Special considerations for those with actionable driver mutations' above.)

●Most studies indicate that stage-matched survival is worse for LCNEC compared with NSCLC and other large cell carcinomas, and similar to that of SCLC. (See 'Prognosis' above.)

ACKNOWLEDGMENT — The UpToDate editorial staff acknowledges Bonnie S Glisson, MD, FACP, who contributed to an earlier version of this topic review.