Tumor Mutation Burden May be a Prognostic Biomarker of Long-Term Survival in Resected Small-Cell Lung Cancer

Small cell lung cancer (SCLC) has a poor prognosis. The majority of SCLC patients do not survive within a few months of diagnosis, however, a sub-group of patients have a long-term survival. Individual differences in prognosis remain elusive. We present the rst comprehensive comparative genomic proling and tumor mutation burden (TMB) analyses of SCLC on patients with long-term survival (LTS) and short-term survival (STS) after surgery. The present study included 52 patients with SCLC who underwent surgery in Zhejiang Cancer Hospital from April 2008 to December 2017. A total of 6 LTS patients ( ≥ 4 years) with stage IIB or IIIA SCLC and 5 STS patients (<2 years) with stage IA or IB SCLC were included. The latter subjects were used as control subjects. All subjects underwent resection without neoadjuvant therapy. We assessed their genomic prole and calculated TMB using next-generation sequencing (NGS). Moreover, we assessed the correlation between TMB and prognosis. Subsequently, we analyzed and compared the molecular characteristics of LTS and STS.


Abstract
Background Small cell lung cancer (SCLC) has a poor prognosis. The majority of SCLC patients do not survive within a few months of diagnosis, however, a sub-group of patients have a long-term survival. Individual differences in prognosis remain elusive. We present the rst comprehensive comparative genomic pro ling and tumor mutation burden (TMB) analyses of SCLC on patients with long-term survival (LTS) and short-term survival (STS) after surgery.

Methods
The present study included 52 patients with SCLC who underwent surgery in Zhejiang Cancer Hospital from April 2008 to December 2017. A total of 6 LTS patients (≥4 years) with stage IIB or IIIA SCLC and 5 STS patients (<2 years) with stage IA or IB SCLC were included. The latter subjects were used as control subjects. All subjects underwent resection without neoadjuvant therapy. We assessed their genomic pro le and calculated TMB using next-generation sequencing (NGS). Moreover, we assessed the correlation between TMB and prognosis. Subsequently, we analyzed and compared the molecular characteristics of LTS and STS.

Conclusions
High nonsynonymous TMB was associated with improved prognosis on patients with resected SCLC. The FAT3 gene may impact disease prognosis. The data may provide valuable information of differences between individuals in terms of prognosis and guide treatment. Studies involving larger groups are required to con rm these ndings.

Page 3/12
Background Lung cancer remains the leading cause of cancer mortality worldwide and in China [1,2]. Small cell lung cancer (SCLC) nearly makes up 15% of all lung cancers, and it is recognized as a highly aggressive and lethal malignancy often seen with early development, extensive metastasis and rapid recurrence following treatment. SCLC patients can be divided into limited diseases (LD) and extensive diseases (ED) with clinical stage. LD patients have a chance being cured with a long-term survival rate of 20-25% when treated with the standard chemoradiotherapy and prophylactic cranial irradiation (PCI). ED patients are theoretically incurable. The median overall survivals (OS) for patients with LD and ED are approximately 15-20 months and 8-10 months, respectively. The median OS is recorded to range from 29 to 91 months among patients who take comprehensive treatment including surgery [3][4][5][6], with an extended therapeutic window, which is worthy of consideration.
Despite the existence of rare long-term survivors of SCLC, there is still relatively a lack of prognostic factors apart from the disease staging and the performance status (PS). Furthermore, it is di cult to explain the ability of certain patients in late stage to get long-term survival, as well as the ability of patients in early stage to relapse among the patients having optimal PS scores. Consequently, in the absence of an adjusted multivariate analysis, it is di cult to assess the prognostic signi cance of the reported biomarkers beyond clinical variables. However, the determinants of long-term survival in SCLC remain largely unknown. Comprehensive genomic pro ling is probably required to establish more robust prognostic markers in SCLCs.
In order to explore the potential genetic alterations and identify prognostic biomarkers in long-term survival (LTS) we analyzed a case control study comparing surgically resected tumors of late-stage ( or IIB) long-term survival with tumors of early-stage (I) short-term survival (STS). Here, we investigate the genomic pro ling and tumor mutation burden (TMB) of LTS and STS using a gene panel (OrigiMed, Shanghai, China) in resected SCLCs, covering all the coding exons of 450 cancer-related genes and 64 selected introns of 39 genes that are frequently rearranged in solid tumors. We expect to nd the genetic differences that represent a contributor to survival.

Sample collection
Postoperative tissue specimens were retrospectively obtained from 52 patients with SCLC. All included 52 patients who underwent surgery at Zhejiang Cancer Hospital (Hangzhou, China) between April 2008 to December 2017 [7] in view of at least 2-year follow-up. All patients were diagnosed with conventional SCLC and the pathological diag nosis was based on the standard criteria de ned by the World Health Organization classi cation [8]. Tumor stage was de ned according to the lung cancer TNM classi cation, eighth edition [9].
The selection process of the SCLC patients is demonstrated in Figure 1. Patients who experienced neoadjuvant chemotherapy or chemoradiotherapy were excluded from our research due to potential puzzling effects of treatment-induced DNA damage. The two special subsets were selected from 52 patients. Patients screened in the study based on the following selection criteria: LTS: Stage or IIB and OS≥4 years, su cient tumor tissue for testing; STS: Stage and OS<2 years, su cient tumor tissue for testing. A total of 6 subjects met the standard of LTS and 5 that of STS (Fig. 1). The present research was approved by the Medical Ethics Committee of Zhejiang Cancer Hospital. The overwhelming majority of the specimens in the study were obtained from the Biological Sample Bank of Zhejiang Cancer Hospital, and the patients signed the written informed consents to have their specimens preserved in the Biological Sample Bank of Zhejiang Cancer Hospital for use in the research study. A limited number of patients were deceased.

Sample preparation
From each tumor-rich formalin-xed para n-embedded (FFPE) and matched normal lung tissue block, 4 μm of sections were cut, depara nized and dissected to isolate 1 cm² of tumor tissue. DNA was isolated using the Cobas R DNA Sample Preparation Kit, according to the manufacturer protocol (Roche Molecular Systems, Pleasanton, CA, USA). The DsDNA concentration was determined using the Qubit R _ 2.0 Fluorometer and the Qubit R _ 2.0 dsDNA HS Assay Kit (ThermoFisherScienti c, Waltham, MA, USA). The quality of the sample DNA was evaluated using a specimen control size ladder test (Invivoscribe Technologies, San Diego, CA, USA).

Next-generation sequencing
The genomic information was produced by NGS-based YuanSu TM 450 gene panel (OrigiMed, Shanghai, China) which covers all the coding exons of 450 cancer-related genes and 64 selected introns of 39 genes that are frequently rearranged in solid tumors. The genes were captured and sequenced with a mean depth of 800X, using Illumina NextSeq 500 (Illumina, Inc). Genomic alterations (GAs) were identi ed by the alignment of sequences from tissues and matched normal lung tissue, following the previously reported methods [10]. Tumor mutation burden (TMB) was estimated by counting the somatic mutations, containing SNVs and Indels, per megabase of the sequence examined on each patient. The driver mutations and recorded germline alterations were not counted.

Statistical analysis
Statistical analyses were performed with the SPSS version 22.0 (SPSS Inc). The signi cance of the differences was analyzed with the Fisher's exact test. A P value lower than 0.05 (p<0.05) was considered to indicate a signi cant difference.

Follow-up
The follow-up deadline was March 17, 2020. Five patients are still alive and no patient was lost to followup and six patients were deceased. The survival time was counted from the date of pathological diagnosis.

Patient characteristics
The median overall survival (mOS) was 87 months (range: 51-143) in the LTS group and 18 months (range: [16][17][18][19][20][21][22] in the STS group. The patient characteristics are presented in table 1. The LTS group included 3 male and 3 female subjects, whereas the STS (n=5) patients were all male subjects. Patients in the LTS group aged from 49 to 63 years old (57 ± 6), while that of STS ranged from 38 to 76 years old (57.8±15 years old). Four patients were heavy smokers (≥20 packs/year) and the 5 remaining patients did not have a smoking history ( Table 1).

The association between TMB and prognosis
The median TMB rates of the LTS and STS groups were 16.4 Mutations/Mb (19.25±9.48) and 8.5 Mutations/Mb (9.88±5.35) respectively. We de ned 10 mutations/Mb as the cut-off value. A value higher than 10 (TMB>10 mutations/Mb) was considered high and a value lower than and/or equal to 10 (TMB≤10 mutations/Mb) low. The differences between the LTS and STS groups were assessed via the Fisher's exact test. The P-value was 0.08 ( Figure 2). Moreover, univariate analysis indicated signi cantly longer overall survival (OS) with high TMB than that with low TMB (NA vs. 22 months, p=0.007) ( Figure   3). Univariate analysis indicated that high TMB was an independent prognostic factor for OS with adjustment for age, sex, and smoking status ( Table 2). The clinical and TMB characteristics are provided for each patient in Supplementary Table S1.

Page 6/12
The research aimed to investigate whether genomic alterations in SCLC may differentiate LTS and STS who experienced surgical resection for early-stage SCLC. The data demonstrated that high TMB may be a prognostic biomarker of long-term survival in resected SCLC, regardless of the disease stage. A tendency was noted for the FAT3 gene to separate between the LTS and STS groups. Although the present study contains a limited number of samples, our data suggests a discovery that warrants further investigation.
The association between TMB and disease prognosis in lung cancer is still unclear. Whether TMB is a prognostic factor for postoperative non-small-cell lung cancer (NSCLC) is controversial. Devarakonda et al. proved that high nonsynonymous TMB was signi cantly connected with a favorable prognosis in resected NSCLC [11]. In contrast to this study, Owada-Ozaki et al. demonstrated opposite results suggesting that high TMB could be related to a poor prognostic in lung cancer and in resected NSCLC [12]. However, the former study contained 908 samples, while the latter only 90 samples. As a consequence, the former conclusion seems more convincing. McGranahan et al. observed an association between higher OS and high neo-epitope burden on early-stage lung adenocarcinoma patients [13]. Yu et al. assessed 255 specimens from patients suffering early-stage squamous cell lung cancers (SqCLC) to evaluate the potential role of PD-L1 protein expression and TMB, and the putative identi cation of an immune gene signature [14]. The authors of that study found that TMB was not associated with OS. SCLC is characterized by high TMB because of its association with smoking. However, whether TMB affects the prognosis of resected SCLC has not been studied. Zhou et al analyzed the correlations between clinical outcomes and genomic alterations in 53 SCLC samples [15]. The authors of this study reported that high TMB (> 21 mutations/Mb) was connected with a favorable prognosis in OS (21.7 vs. 10.4 months, p = .012). In our study, we found that LTS exhibited higher median TMB (16.4 mutations/Mb, ranging from 10.8 to 35.6 compared with 8.5, from 5.2 to 17.9). There was no signi cant difference in the median TMB between the LTS and STS groups (p = 0.08) (Fig. 2). However, exploratory analyses suggested that TMB may have a signi cant predictive effect on OS (p = 0.007) (Fig. 3). For patients with high TMB and low TMB, The former group did not reach mOS and the latter one survived 22 months. These ndings provide evidence that high TMB exhibits optimal prognostic value.
The genetic mutational landscape of SCLC is complex and heterogeneous. However, the most common genetic alterations include inactivation of the tumor suppressor genes TP53 and RB1 [16,17]. Hu et al. demonstrated that the most frequently altered genes in small cell lung cancer were as followed: TP53 (93.4%), RB1 (78.7%), LRP1B(18.9%), KMT2D (15.6%), FAT1 (11.5%), KMT2C (11.5%),STK24 (11.5%), FAM135B (10.7%), and NOTCH1 (10.7%) [10]. The results were derived from genomic pro ling of 122 Chinese patients. In the study, with the exception of TP53 and RB1, the remaining high frequency mutated genes were those that encoded enzymes involved in histone modi cation, notably those that participate in the NOTCH and Wnt signaling pathways. With the Fisher's exact test, we found a signi cant correlation between LTS and gene mutations on FAT3. The human FAT gene family consists of the FAT1, FAT2, FAT3 and FAT4 genes [18][19][20][21]. Hong et al reported the partial coding sequence of FAT3 in 2004, whereas Katoh et al reported the complete coding sequence of FAT3 and FAT4 in 2006. FAT3 is found on chromosome 11q14.3-q21 and has 26 exons encoding a protein of 4,557 amino acids [22]. The FAT1 and FAT3 genes adjoin the MTNR1A and MTNR1B genes, respectively. FAT1 exhibits a higher homology with FAT3, while MTNR1A exhibits a higher homology with MTNR1B. A previous study in mice indicated that FAT3 expression restricted the development of central nervous system (CNS), with highest expression found at the olfactory bulb and retina [23]. These ndings led to the hypothesis that murine FAT3 plays signi cant roles in the development of CNS, possibly in axon organization and interaction [23,24]. Sadeqzadeh et al reported that on patients suffering breast and ovarian cancer exhibited high frequency FAT3 mutations and FAT3 impacts the development of the central nervous system [22,25,26]. Ji-Yeon Kim et al examined 119 patients of breast cancer in an exploratory biomarker study: a total of 40 subjects from that study exhibited an available biomarker. With targeted deep sequencing, FAT3 (48%) was found to be the most frequently mutated gene. However, further survival analysis indicated that FAT3 mutations seemed to be associated with poor prognosis, though the results were statistically insigni cant [27]. The information regarding the association between FAT3 and the prognosis in SCLC was lacking in the present study. Interestingly, FAT3 mutation occurred solely in LTS and the P-value from the comparison with the STS group was 0.06 as determined by the Fisher's exact test. The survival curve analysis (Fig. 3) also showed that FAT3 mutation may be associated with optimal prognosis among SCLC patients, although there was no notable difference between them (p = 0.11) (Fig. 3). This may be the rst report regarding the effects of FAT3 mutations on the prognosis of SCLC patients. However, the reports that exist on FAT3 are limited and it remains unknown whether the genetic changes occurring in FAT3 can signi cantly affect the pathophysiology of the cancer [22]. Further studies that can prove the implication of FAT3 in SCLC would be of signi cant value.
This study has a few limitations. The sample size was small and this was retrospective study. Surgical specimens were scarce and precious due to the less opportunity for surgery. For this reason, scienti c studies about SCLC molecular pro les are hampered by a lack of tissue availability. Moreover, in order to thoroughly ignore the effect of stage on prognosis, we selected two extreme cohorts of LTS with stage IIB-III and STS with stage I as the objects of study. All the patients were followed up for more than 2 years. Therefore, the number of patients who quali ed is very small. Despite these limitations, our study offered some new discoveries.

Conclusions
We studied the genomic alteration pro le and TMB of LTS and STS. High nonsynonymous TMB may be considered an optimal prognostic biomarker of long-term survival in resected SCLC. The FAT3 gene may exert an impact on prognosis and it may be a potential and interesting gene to study. Further studies are warranted to con rm these observations and explore the mechanisms underlying this association.

Declarations
Ethics approval and consent to participate: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and national research committee and with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. The Medical Ethical Committee of Zhejiang Cancer Hospital approved this study. This study is a retrospective study, and the overwhelming majority of the specimens in the study were obtained from the Biological Sample Bank of Zhejiang Cancer Hospital, and the patients signed the written informed consents to have their specimens preserved in the Biological Sample Bank of Zhejiang Cancer Hospital for use in the research study.