Combined SCLC and NSCLC histology

Treatment strategies for lung cancer are based on the assumption that an individual patient's cancer is purely of one subtype.^11,12 For example, limited-stage SCLC and localised NSCLC are both potentially curable diseases, but their treatment strategies differ substantially.^11,13,14 Surgery has been largely abandoned as a treatment modality in patients with limited-stage SCLC, which is now treated with concurrent chemotherapy and radiotherapy.¹¹ By contrast, patients with localised stage I or II NSCLC are treated mainly with surgery possibly followed by adjuvant chemotherapy, although radiotherapy is typically not part of the strategy.¹⁵ In advanced stage lung cancer, chemo therapy is the mainstay of treatment for both NSCLC and SCLC but the drugs used are distinct for each subtype. The initial response to chemotherapy is much greater for patients with extensive-stage SCLC than patients with metastatic adeno carcinoma,^11,12 which suggests inherent differences in tumour biology. For example, pemetrexed is one of the most active agents in meta static adeno-carcinoma,¹² but it is fairly inactive in extensive-stage SCLC.¹⁶ This difference could result from the high levels of thymidylate synthase in patients with SCLC,¹⁷ which can cause resistance to pemetrexed.¹⁸

In practice, combined-histology lung cancers can make treatment decisions difficult. SCLC with a large-cell component accounts for about 10% of SCLC cases; at present, these patients are given standard SCLC chemo-therapy regimens.^11,14 Two large case series have investigated the frequency of tumours with combined SCLC and NSCLC histology.^19,20 In the first study,¹⁹ 176 SCLC tumours were analysed and 17 (10%) tumours also had an NSCLC component. Seven of these patients who had originally been diagnosed with NSCLC and given chemotherapy or radiotherapy were found to have combined histology at surgery or autopsy. Additionally, in autopsies of 40 patients originally diagnosed with SCLC, two were reported to have only adenocarcinoma. In the second case series,²⁰ the histology of 429 SCLC tumours was studied; nine (2%) tumours contained NSCLC (six adenocarcinoma, three squamous-cell carcinoma). Repeat biopsy samples from six of these nine patients showed one dominant histology (three SCLC, two adenocarcinoma, one squamous-cell carcinoma). Although one possible ex planation for these results is the transformation from one phenotype to another, another possibility is that core biopsy samples or fine-needle aspirates used to make the initial diagnosis did not provide sufficient pathological material to determine the presence of combined histology to be identified at diagnosis. Whether the results from these autopsy studies represent transformation or com bined-histology tumours, these striking observations pose several practical questions that could affect clinical practice. First, in localised, early-stage, combined-histology tumours, how does the presence of adenocarcinoma affect the decision on whether to offer surgery? Alternatively, should patients with tumours that have an SCLC component always be given concurrent chemotherapy and radio therapy? Furthermore, in an advanced-stage setting, if tumours do not respond as initially expected or if the responses among two or more different lesions are discordant, is repeat biopsy indicated to rule out a dominant histology that was not identified at initial diagnosis?

EGFR-mutant adenocarcinoma and transformation to SCLC

Three EGFR inhibitors—erlotinib, gefitinib, and afatinib—are now used worldwide for first-line treatment of lung cancers that have EGFR-activating mutations, and their use has greatly changed clinical practice. However, within an average time of 12 months,⁷ resistance develops, and samples from repeat biopsies have shown several distinct mechanisms of acquired resistance to EGFR inhibitors. The most common resistance mechanism is a Thr790Met mutation in EGFR, which increases the affinity of the receptor for ATP and allows for continued EGFR signalling in the presence of the inhibitor.²¹ Thr790Met mutations are reported in 50–60% of samples from resistant tumours.^9,10 Mechanisms that bypass the requirement for EGFR signalling, such as MET and HER2 amplification, could account for another 15–20% of resistance to EGFR inhibitors.^9,10,22

Another mechanism of resistance to EGFR inhibitors is histological transformation of EGFR-mutant adeno-carcinoma to SCLC. This process was first described in 2006 in a 45-year-old woman with EGFR-mutant adenocarcinoma who had never smoked; she was originally diagnosed with adenocarcinoma and given erlotinib.²³ The tumour responded for 18 months, but at the time of progression, the repeat biopsy sample showed SCLC with the original EGFR exon 19 deletion mutation. Since this first case, several other case series have been reported of EGFR-mutant adenocarcinoma transforming to SCLC as a mechanism of resistance to EGFR-inhibitor treatment (table 1).^9,10,23–28 In all reported cases, small-cell tumours were identified by morphology and positive immunohistochemical staining for synaptophysin, chromogranin, or NCAM. In a case series in which repeat biopsies were done when resistance developed in patients given EGFR inhibitors,⁹ transformation to SCLC was reported in 14% of cases. In another case series, trans formation to SCLC was apparent in less than 5% of repeat biopsy samples.¹⁰ Genomic sequencing of EGFR from both the original and repeat biopsy samples at the time of resistance shows that every transformed SCLC tumour sample retained its original EGFR-activating mutation (table 1),⁹ which suggests that these were not independent de-novo cancers, but a transformed phenotype as a mechanism of resistance to treatment instead. Another hypothesis for this observation is that patients had tumours with combined histology at the time of initial diagnosis, which was not apparent on the diagnostic biopsy sample, and then the SCLC component became dominant as the adenocarcinoma component was successfully treated with the EGFR inhibitor. However, the clinical course of the reported cases has not typically been consistent with this hypothesis because most patients responded well to EGFR inhibitors for several months to years (table 1)⁹ and clinically had greater tumour growth at the time when SCLC was diagnosed. If tumour heterogeneity accounted for the development of SCLC in these patients, a less dramatic response to EGFR inhibitors and acquired resistance at an earlier time would be expected.

Does transformation to SCLC result solely from EGFR inhibition, or does EGFR-mutant adenocarcinoma have the potential to transform into SCLC? Furthermore, are adenocarcinomas that do not have EGFR mutations similarly able to transform to SCLC? Synchronous development of adenocarcinoma and SCLC was noted in EGFR-mutant tumours before treatment with an EGFR inhibitor.²⁶ This observation suggests that the development of SCLC in EGFR-mutant cancers is not exclusively the result of EGFR inhibition. Furthermore, this series²⁶ included two cases of adenocarcinoma that transformed to SCLC in tumours that did not have EGFR driver mutations in the original biopsy sample; this finding suggests that transformation can occur independently of EGFR mutational status. This was unlikely to be the result of false-negative EGFR testing because in both cases the specimens tested were resected tumours and one patient was reported to have a KRAS mutation. In a case series reported in 1986, before the discovery of EGFR-activating mutations, when resistance to conventional chemotherapy or radiotherapy developed, about 5% of patients originally diagnosed with NSCLC had SCLC at relapse.¹⁹ Whether the tumours of these patients did have EGFR-activating mutations is unknown, but the patients did show transformation to SCLC independently of EGFR inhibition. In their 2011 study, Sequist and colleagues⁹ found no cases of transformation to SCLC in 79 surgical specimens from patients with stage III NSCLC who received chemotherapy and radiotherapy,⁹ which suggests that EGFR-wild-type NSCLC is less likely to transform to SCLC than EGFR-mutant NSCLC. Larger studies will be needed to identify the precise frequencies of SCLC transformation in EGFR-mutant and EGFR-wild-type NSCLC, but the information available at present suggests that transformation to SCLC is more common in EGFR-mutant cancers treated with EGFR inhibitors than in wild-type cancers.

EGFR mutations in de-novo SCLC

Two large case series^29,30 have investigated whether EGFR mutations occur de novo in classic SCLC (table 2).^26,29–35 In the series reported by Tatematsu and colleagues,³⁰ five of 122 patients had EGFR-mutated SCLC. These patients had a history of heavy smoking (average of 30 pack-years), but a significantly lower pack-year smoking history than that of patients who did not have EGFR mutations. Three of the five patients with EGFR-mutant SCLC had combined small-cell and adeno carcinoma histology. Two of the patients, both of whom had EGFR amplification, were given gefitinib and responded to treatment. One responder had combined SCLC and adenocarcinoma with EGFR amplification reported only in the adenocarcinoma component. The other responder had small-cell histology with EGFR amplification and had a rare Gly719Ala EGFR mutation. Of note is another case report of a patient with de-novo SCLC with an EGFR-activating mutation that did respond to gefitinib.³¹ Immunohistochemistry showed that this patient had high expression of EGFR. In another large case series from Taiwan,²⁹ 76 patients with SCLC were analysed for the presence of EGFR mutations. Two patients had EGFR mutations, who did not have previously known EGFR-mutant adenocarcinoma. One patient was treated with gefitinib and showed no response to treatment.²⁹

In a phase 2 trial,³⁶ responses to gefitinib were tested in 19 patients with SCLC previously given chemotherapy. The participants were not tested specifically for EGFR mutations.^29,30 Only two of the patients had stable disease and there were no partial or complete responses.³⁶ Analysis of EGFR-mutant tumours that transformed to SCLC from adenocarcinoma as a mechanism of acquired resistance shows that, at transformation, the cancers lose protein expression of EGFR and have lower levels of EGFR amplification.³⁷ In the past, SCLC was reported to have lower protein expression of EGFR than NSCLC.^38,39 The mechanisms underlying low expression of EGFR in SCLC remain unknown, but this could explain the absence of response to EGFR inhibitors in several case reports.^23,28,29 Overall, response to EGFR inhibitors in SCLC patients with EGFR mutations do not seem to match those of EGFR-mutant adenocarcinoma, perhaps because of mechanisms that suppress EGFR protein expression in these tumours. However, a case report by Araki and colleagues³¹ suggests that patients can have EGFR-mutant SCLC that does express EGFR and responds to EGFR inhibition. In view of individual case reports of patients with rare SCLC with EGFR amplification, larger studies will be needed to correlate protein expression with EGFR inhibition to find out whether other patients with SCLC could benefit from EGFR inhibition.

Genomic analysis of SCLC and adenocarcinoma

Genomic analyses have improved understanding of the pathways that are dysregulated in each specific tumour type. These efforts aim to identify oncogenic driver mutations, copy number changes, and translocations that can be targeted therapeutically. In adenocarcinoma, these sequencing efforts have resulted in the identification of several mutated oncogenic drivers and tumour suppressor genes and have validated the high prevalence of KRAS and EGFR mutations.⁴⁰ In clinical practice, the standard of care is to genotype every patient with a new diagnosis of metastatic adenocarcinoma for EGFR mutations and ALK translocations, since targeted therapies are now approved in cancers that have these abnormalities.^7,41,42 EGFR mutations and ALK translocations are estimated to account for about 30–50% of adenocarcinomas in patients who have never smoked, and the proportion might vary according to race.^43,44 Additionally, many centres now undertake multiplexed genetic analyses to identify the presence of other prevalent genetic abnormalities, since therapeutics are under development for those cancers also.⁴⁵

Two large-scale genome-sequencing projects have been completed in SCLC.^46,47 They included whole-genome, transcriptome, exome, and copy-number analyses. Both studies identified a high prevalence of mutations in TP53 and RB1, validating previous human sequencing studies. Indeed, mouse models of SCLC have been developed by conditional inactivation of Tp53 and Rb1.⁴⁸ Although some findings overlapped between these two studies, several findings were unique to each study, perhaps because of the heterogeneous patient samples in terms of stage and exposure to previous treatment. Known oncogenic drivers reported in adenocarcinoma, such as EGFR or KRAS, were not statistically significantly mutated in either SCLC genome-sequencing study. Of note, MYC amplification was seen in 16% of cases.⁴⁶ MYCL1 knock-down resulted in decreased cell proliferation in SCLC cell lines,⁴⁷ which suggests that MYC functions as an oncogenic driver in a subset of small-cell tumours. Activators of ERK signalling, such as EGFR and KRAS, are mutated more commonly in adenocarcinoma. By contrast, RB1 loss seems to be much more common in SCLC.^46–48 Together these genomic efforts suggest that fundamental genetic differences drive the development of SCLC and lung adenocarcinoma.

RB1 inactivation as a defining feature of SCLC

Inactivation of RB1 has long been known to have an important role in the tumorigenesis of SCLC. In 2003, a mouse model of SCLC was developed by means of conditional knockout of Tp53 and Rb1.⁴⁸ Further studies with this model have shown that Rb1 inactivation is essential for the development of SCLC. Mutation or loss of RB1 has been found in 100% of the human SCLC tumours sequenced.⁴⁶ Furthermore, a proteomic study that compared SCLC and NSCLC cell lines showed that RB1 was quantitatively the most downregulated protein in SCLC compared with NSCLC.¹⁷ However, mice with loss of Rb1 alone do not develop small-cell tumours, which suggests that loss of Rb1 is necessary but not sufficient for SCLC tumorigenesis. Analyses of repeat biopsy samples from patients with EGFR-mutant adenocarcinoma that transform to SCLC revealed that 100% of these patients have loss of RB1;³⁷ loss of RB1 seems to be a universal event leading to transformation from adenocarcinoma to SCLC.

Prostate cancer is another example of a cancer in which small-cell differentiation can occur after treatment. In one case series, 90% of small-cell prostate tumours had loss of RB1.⁴⁹ Furthermore, loss of RB1 selectively occurred in castration-resistant prostate cancer, which suggests that loss of RB1 is a late event in tumorigenesis.⁵⁰ RB1 loss is well known to be frequent in many non-neuroendocrine tumour types also.⁵¹ Why neuroendocrine differentiation occurs only in some tumour types with loss of RB1 is not known. Evidence from animal models supports a role for loss of Rb1 selectively affecting tumours of neuroendocrine origin. For example, Rb1+/− mice form pituitary and thyroid tumours after loss of the wild-type allele.⁵² Additionally, conditional inactivation of Rb1 in lung tissue results in increased numbers of pulmonary neuroendocrine cells during development.⁵³ The three cancers with the highest frequency of RB1 mutations or deletions are SCLC, prostate cancer, and bladder cancer.⁵¹ Interestingly, both prostate and bladder cancers are known to undergo small-cell differentiation.^49,54 Additional mechanistic insight into the role of RB1 loss in transformation to SCLC will help elucidate the biology of SCLC, how transformation from adenocarcinoma to SCLC occurs, and possibly help create new therapeutic targets to treat SCLC.

This work was funded by R01 CA137008 (JAE) and T32 CA009172 (MGO).