YAP1 Protein Expression Has Variant Prognostic Signicance in SCLC Stratied by Histological Subtypes

Purpose: Yes-associated protein 1 (YAP1), a nuclear effector of an inactivated HIPPO pathway, has been identied as one of four molecular subtypes of SCLC dened by the transcriptional regulatory mechanism. We aimed to explore the clinicopathological relevance and prognostic signicance of YAP1 expression in SCLC stratied by histological subtypes. Methods: Tumor sections and corresponding formalin-xed paran-embedded (FFPE) samples of 297 SCLC patients were retrieved from the pathological specimen repository and were subsequently reviewed by pathologists. Forty-six C-SCLCs (15.5%) and 251 P-SCLCs (84.5%) were identied respectively. YAP1 expression was examined by immunohistochemistry (IHC) and assessed semi-quantitatively on tumor tissue array (TMA). Propensity score was used to match C-SCLCs and P-SCLCs in a ratio of 1 to 2 to balance age, gender, tumor stage and treatment methods. Finally, 46 C-SCLCs and 92 P-SCLCs were included for prognostic analysis. Results: The positive rate of YAP1 expression was signicantly higher in C-SCLCs than P-SCLCs before matching (52.2% vs 29.1%, P = 0.004). After matching by propensity score, the prescribed clinical parameters were well balanced between P-SCLCs and C-SCLCs. Expression of YAP1 was associated worse overall survival (OS) (5- year OS%, 39.0% vs. 74.9%, P = 0.013) and was an independent risk factor for OS (HR = 2.93, 95% CI: 1.01-8.51; P=0.048) exclusively in C-SCLC. Univariate survival analysis in subgroups of different clinical variables also conrmed the prognostic impact of YAP1 was most signicant in C-SCLC. But for P-SCLCs, expression of YAP1 showed no prognostic impact. Conclusions: Expression of YAP1 in small cell components of C-SCLC was signicantly higher than that in P-SCLC. Besides, it served as an unfavorable predictor for OS in C-SCLC but not in P-SCLC, which suggested different entities of small cell components with variant YAP1 expression and potential different targetable oncogenic pathway between C-SCLC and P-SCLC.


Introduction
Lung cancer is one of the most lethal malignancies in China, (Chen et al. 2016) of which small cell lung cancer (SCLC), the most aggressive histological type of lung cancer, accounts for 15% ∼ 20%. ( (pure SCLC, P-SCLC), while some can be combined with additional components of any histological types of non-small cell lung cancer (NSCLC), which is de ned as combined small cell lung cancer (combined SCLC, C-SCLC). (Brambilla et al. 2001) The reported prevalence of C-SCLC is variable, ranging from 2-28% of all SCLC cases in different studies, which might be in uenced by different methods for sampling. (Babakoohi et al. 2013; Mangum et al. 1989; Nicholson et al. 2002) Previous studies have shown that the clinical characteristics of C-SCLC do not differ signi cantly from those in patients with P-SCLC. (Guo et al. 2020;Mangum et al. 1989; Nicholson et al. 2002) In addition, C-SCLCs have been treated based on SCLC guidelines, (Moon et al. 2019) similarly with a platinum agent and etoposide with or without thoracic radiation, after adjustments for tumor extent or stage for over three decades. (Gazdar et al. 2017) Therefore, to date, SCLC has been regarded as a 'homogenous' disease with little documented inter-tumor heterogeneity with respect to histology or molecular biology in clinical practice. (Gazdar et al. 2017) However, it has been reported that C-SCLC patients have lower response rates to chemotherapy compared to P-SCLC patients in some studies. (Hirsch et al. 1988; Radice et al. 1982) These results indicated histological subtypes have an impact on clinical management of SCLC and we can no longer regard SCLC as a single disease entity. Besides, the difference of molecular biological characteristics between different histological subtypes in SCLC is worth further exploring to guide more effective and individualized therapy.
Recently, expression of YAP1 together with achaete-scute family bHLH transcription factor 1 (ASCL1), neuronal differentiation 1 (NEUROD1) and POU class 2 homeobox 3 (POU2F3) was identi ed as one of four molecular subtypes of SCLC de ned by the predominant transcriptional regulatory mechanism operating in the tumor cells. However, the clinicopathological relevance and prognostic signi cance of YAP1 expression in SCLC has not been systematically reported to date. YAP1 is a downstream nuclear effector of the inactivated Hippo signaling pathway, which is essential for regulating cell proliferation, apoptosis, stem/progenitor cell expansion and organ growth and is considered as a conserved tumor suppressor pathway. ( 2009) However, its tumor-suppressive roles were also demonstrated in some cancers such as breast cancer, hematological malignancies (leukemia and multiple myeloma) and SCLC. ( However, in SCLC, the impact of YAP1 expression on tumor chemosensitivity seems elusive and paradoxical. One study showed that SCLC cell lines with high YAP1 expression were more resistant to cisplatin and among patients receiving adjuvant chemotherapy, YAP1 positive cases had a signi cant shorter survival than YAP1 negative cases, but for patients without adjuvant chemotherapy, YAP1 positive patients had a better prognosis, (Ito et al. 2016) suggesting potential co-existing roles as a tumor suppressor and a drug-resistant molecule. Nevertheless, another study which analyzed the data on the response to 526 anticancer agents in 61 SCLC cell lines showed SCLC cell lines with high YAP1 expression had higher sensitivity to etoposide and topotecan which were widely used for SCLC and were more sensitive to mechanistic target of rapamycin kinase (mTOR) and polo like kinase (PLK) inhibitors. Therefore, the author thought SCLC of the YAP subgroup may be more responsive to chemotherapy or targeted therapies. (Horie et al. 2016) Due to the complex signal network YAP1 involved in and dual role either as a tumor promoter or a tumor suppressor, the clinicopathological relevance of YAP1 in SCLC remain to be investigated. Besides, C-SCLC is a more biologically complex tumor consisting of both small cell and nonsmall cell components, in which YAP1 may have different impacts. Our study compared the expression of YAP1 between P-SCLC and C-SCLC mainly in the small cell components and explore its clinicopathological relevance respectively, aiming at demonstrating the difference of molecular biological characteristics and its impact on treatment and prognosis between P-SCLC and C-SCLC.

Patient selection and histological identi cation
Three hundred and forty-three patients who received a surgery and diagnosed as SCLC in Cancer Hospital, Chinese Academy of Medical Science (CHCAMS) between 2005 and 2016 were initially included. Clinical data including age, gender, smoking history, tumor laterality, TNM stage, stage by Veterans Administration Lung Study Group (VALSG), operation means and post-surgery treatment methods were extracted from medical record system. Forty-six patients without complete follow-up information were excluded. Tumor sections and corresponding formalin-xed para n-embedded (FFPE) samples of the remaining 297 SCLC patients with complete medical records and follow-up information were retrieved from the pathological specimen repository of CHCAMS and were subsequently reviewed by one senior pathologist (Lin Yang) and two junior pathologists (Li Liu and Xin Wang). And the histological subtype was identi ed and con rmed independently according to the 2015 World Health Organization classi cation of lung tumors. (Travis et al. 2015) As for the combined pathology, we considered presence of at least 10% large cell for combined small cell and large cell carcinoma diagnosis. For small cell/squamous carcinoma or small cell/adenocarcinoma, presence of any amount of non-small-cell component was considered diagnostic. (Nicholson et al. 2002) Based on this criterion, 46 and 251 samples were identi ed as C-SCLC and P-SCLC respectively. We used propensity score to match C-SCLCs and P-SCLCs in a ratio of 1 to 2 with a caliper 0.2 to balance age, gender, tumor stage and treatment methods. Finally, 46 C-SCLCs and 92 P-SCLCs were included for analysis of the prognostic signi cance of YAP1 expression in different histological subtypes (P-SCLC vs. C-SCLC). Propensity score matching (PSM) was performed by R software (version 3.6.3). This study was approved by the Ethics Committee and Institutional Review Boards of CHCAMS (No.20/234-2430) and all patients were exempt from an informed consent due to the retrospective nature of the study.
Tumor section review and evaluation of pathological characteristics Besides the histology identi cation, we further observed and recorded other pathological characteristics including status of lymph nodes (positive or negative), presence or absence of pleural invasion, bronchus invasion, vascular invasion, nerve invasion, tumor thrombosis and phenomenon of spread through air spaces (STAS), and proportion of necrosis, brosis and tumor in ltrating lymphocytes (TILs). The proportions of brosis/necrosis/TILs were de ned as the ratio of the area of brosis or necrosis to the total area of the tumor section. The proportion of brosis or necrosis was de ned as the ratio of the area of brosis or necrosis to the total area of the tumor section. And the denominator used to determine the percentages of TILs is the area of tumor parenchyma. The proportions of TILs were based on a full assessment of average area of TILs within the tumor area from all selected sections as a previous study described. (Liu et al. 2020) Tissue array construction and immunohistochemical (IHC) staining The archived para n blocks corresponding to 297 SCLC cases included were retrieved and constructed into tissue microarrays (TMAs) (1.5mm x 2 punctures per donor block). Serial 4um tissue sections were cut and stained with hematoxylin and eosin (H&E). Staining intensity of diagnostic markers (CgA, CD56, Syn, Ki-67) for SCLC and (P63, P40, TTF-1, NapsinA) for NSCLC reported on the original pathology report sheet were referred for histological identi cation when morphology was equivocal for diagnosis. The rabbit monoclonal antibody against human YAP1 (Abcam Cat# ab52771, RRID: AB_2219141) was used for immunostaining of YAP1. IHC staining was completed on the fully automatic Roche immunohistochemical instruments (Roche Diagnostics, Shanghai, China) according to the recommended standard protocols. YAP1 expression was scored based on the percentage of cell nuclei that stained positively: less than 10% or no staining, (-); 10-25%, (+); 25%-50%, (++); more than 50%, (+++) by one senior pathologist (Lin Yang) and two junior pathologists (Li Liu and Xin Wang). The expression of YAP1 was de ned as positive when 10% of the tumor cell nuclei or greater were stained (scores 1+, 2 + or 3+) and negative when less than 10% were stained (scores -). The clinical and prognostic information of patients was blinded to the operator who performed the immunostaining and the pathologists who evaulated the staining intensity of YAP1 antibody.

Follow-up strategy
The follow-up was conducted by regular patient visits or telephone calls which began on April 11, 2005 and was completed until February 28, 2019. In general, patients with limited stages were recommended for outpatient review every 3 months in year 1 to 2, every 6 months in year 3, and annually thereafter and patient with extensive stages were recommended for outpatient review every 2 months in year 1, every 3 to 4 months in year 2 to 3, every 6 months in year 4 to 5 and annually thereafter as NCCN guideline instructed.(NCCN Clinical Practice Guidelines in Oncology. Version 3.2020. Small Cell Lung Cancer. 2020) The follow-up periods and intervals were then determined according to tumor status and treatment recommended by physicians. The median follow-up time was 47.2 months for overall cohort, 46.5 months for C-SCLCs and 47.8 months for P-SCLCs respectively. Relapse-free survival (RFS) is de ned as the time durance between the start of surgery and the observation of local recurrence or distant metastasis of the tumor con rmed by imaging or biopsy of metastatic sites. In case of no recurrence during follow-up, the endpoint of RFS is the last follow-up or death. Overall survival (OS) is de ned as the time from the date of surgery to death or the last follow-up (in case of no death). The primary endpoint of this study was OS, and second endpoint was RFS.

Statistical analysis
All statistical analysis was performed on SPSS software version 25.0. For continuous normal distribution variables, the mean ± standard deviation was calculated, and the Student's t test was applied to show the signi cance of difference between groups. For continuous abnormal distribution variables, the median and quartile were calculated, and the Wilcoxon rank sum test was applied to show the signi cance of difference between groups. For categorical variables, the percentage was calculated, and the Fisher's exact test or the Chi-square test was applied to determine the signi cance of difference. The Kaplan-Meier method was used to estimate and compare survival, with the log-rank test applied for signi cance testing. The Cox proportional hazards model was used for multivariate survival analysis. Both the univariate and multivariate analysis were performed in P-SCLCs and C-SCLCs separately. All statistical tests were bilateral and P < 0.05 was considered statistically signi cant. P < 0.1 was considered as a trend.
Clinicopathological characteristic distribution before and after propensity score matching (PSM) We used PSM to match C-SCLCs and P-SCLCs in a ratio of 1 to 2 to balance age, gender, tumor stage and treatment methods. Ninety-two cases of P-SCLC patients were successfully matched with 46 C-SCLC patients and the prescribed clinical parameters were well balanced between P-SCLC and C-SCLC cases.
Before matching, the age of C-SCLC patients was higher than P-SCLC patients (59.65 ± 8.72 vs. 56.56 ± 10.12, P = 0.053), but the imbalance was negligible after matching with mean ± standard deviation of age 59.77 ± 8.11 in P-SCLC patients (P = 0.937). Other clinical parameters like gender, clinical stage, VALSG stage, treatment methods were also well balanced with P > 0.05 ( Table 2). The detailed clinicopathologic features and treatment history of 46 C-SCLC cases and 92 P-SCLC cases were shown in Table 2. However, YAP1 expression level was signi cantly higher in small cell components of C-SCLCs than in that of P-SCLCs before and after PSM (before PSM: median 1 vs. 0, P = 0.00058, Fig. 2a; after PSM: median 1 vs. 0, P = 0.011, Fig. 2b). Positive YAP1 expression was associated a worse prognosis in C-SCLC but not in P-SCLC The follow-up period for the entire cohort after matching ranged from 4.0 to 166.6 months, and the median follow-up time was 47.2 months, 47.8 months, 46.5 months for entire cohort, P-SCLC cohort and C-SCLC cohort respectively. In total, 67 patients (48.6%, including 44 P-SCLCs and 23 C-SCLCs) had relapsed and 52 patients (37.7%, including 34 P-SCLCs and 18 C-SCLCs) deceased at the end of followup. During the follow-up, 27 patients (19.6%, including 18 P-SCLCs and 9 C-SCLCs) were lost. The 5-year RFS rates were 51.9%, 54.0%, 47.6% and the 5-year OS rates were 63.3%, 66.4%, 56.7%, for entire cohort, P-SCLC cohort and C-SCLC respectively. The median RFS was 74.7 months, 85.7 months, 59.3 months and the median OS was 110.6 months, 110.6 months, NA for entire cohort, P-SCLC cohort and C-SCLC cohort respectively. No signi cant RFS (P = 0.901) and OS (P = 0.700) were observed between P-SCLC and C-SCLC ( Figure S2). According to univariate survival analysis in the 92 P-SCLC patients, YAP1 expression not signi cantly correlated with RFS (P = 0.863, Figure S1a) or OS (P = 0.728, Fig. 4a). However, for 46 C-SCLC patients, YAP1 positive patients had a shorter RFS and OS compared with YAP1 negative patients (RFS, 25.2 vs. 97.7 months, P = 0.366, Figure S1b; OS, 38.8 vs. NA months, P = 0.013, Fig. 4b), with the 5year RFS rate 42.3% vs. 54.4% (P = 0.366) and the 5-year OS rate 39.0% vs. 74.9%, P = 0.013) in YAP1 positive and YAP1 negative patients respectively, although the effect of YAP1 expression on RFS was insigni cant (P = 0.366, Figure S1b).
Multivariate analysis identi ed positive YAP1 expression as an independent prognostic factor for C-SCLC Variables including age group, gender, smoking history, TNM stage, VALSG staging, present or absent lymph node invasion, whether or not receiving prophylactic cranial irradiation (PCI) and YAP1 expression (positive or negative) were included in the Cox proportional hazard model for multivariate survival analysis in P-SCLC and C-SCLC separately. Positive YAP1 expression was identi ed as one of three independent risk factors for shorter OS in C-SCLC group (HR 2.93, 95% CI 1.01-8.51, P = 0.048, Fig. 5A). The other two independent factors for OS of C-SCLC were VALSG stage (Extensive stage vs. Limited stage HR 19.91, 95% CI 2.75-144.29, P = 0.003, Fig. 5A) and presence of lymph node invasion (HR 4.09, 95% CI 0.93-17.98, P = 0.062, Fig. 5A). In P-SCLC group, only clinical stage was identi ed as an independent prognostic factor (P = 0.008, Fig. 5B) and YAP1 expression had no prognostic signi cance for OS.

Discussion
In this study, to identify the clinicopathological relevance of YAP1 expression, we compared other clinicopathological parameters between YAP1 (+) patients and YAP1 (-) patients, and found that expression of YAP1 had correlation with histological subtype, age, gender and smoking history in SCLC with higher expression observed in patients with C-SCLC (P = 0.004), male sex (P = 0.007), age > 60 years (P = 0.005) and smoking history (P = 0.042). We further assessed the YAP1 expression in non-small cell components of C-SCLC and found that the YAP1 expression was commonly observed, especially in nonneuroendocrine carcinoma components like ADC (83.3%) and SCC (58.0%), with total positive rate 58.7%.
These results were in accordance with the biological function of YAP1 which involved in inhibiting neuroendocrine differentiation. (Ito et al. 2016) However, the small cell components of C-SCLC were also commonly found positive YAP1 expression and expression of YAP1 was consistent with the non-small cell components and even higher than P-SCLC (52.2% vs. 29.1%, P = 0.004). This is reminiscent of a previous study which found that the histologic components of C-SCLC had high genetic concordance with approximately 75% of somatic mutations shared in small cell and non-small cell components, suggesting a common precursor of both components in C-SCLC. Our study supported this hypothesis since concordant YAP1 expression which was relatively rarely expressed in P-SCLC was observed between the two components with only 3/27 patients found positive YAP1 expressed exclusively in nonsmall cell components. Furthermore, positive expression of YAP1 was associated with worse prognosis in C-SCLC but not in P-SCLC, implicating that the small cell components in C-SCLC and P-SCLC might be of different entities with potential different origins, oncogenic and tumor-promoting mechanism which might be selectively targetable in C-SCLC. In addition, YAP1 expression could serve as a potential candidate biomarker for prognosis prediction exclusively in C-SCLC.
Previous studies have shown that compared with P-SCLC patients, C-SCLC patients have different response rates to chemotherapy indicating underlying different molecular mechanisms between histological subtypes to regulate this response. YAP1 is a molecular involved in drug metabolism and its expression in SCLC cell lines is found in uencing the sensitivity to chemotherapeutic drugs. (Ito et al. 2016;Yimlamai et al. 2015) Besides, in SCLC patients receiving adjuvant chemotherapy, different OS was observed between YAP1 (+) and YAP1 (-) patients. (Shibata et al. 2018) However, in the current study, we found the prognosis was not affected by YAP1 expression in P-SCLC patients. But in C-SCLC, YAP1 expression was indeed found correlated with prognosis since YAP1 (+) C-SCLC patients suffering a signi cant worse OS. Survival analysis in subgroups of different clinical variables also con rmed the prognostic impact of YAP1 was most signi cant in C-SCLC. This phenomenon might have translational signi cance in regarding the YAP1 as a candidate target for treatment of C-SCLC especially for patients with higher expression of this molecule. Since YAP1 is a vital downstream transcription factor in an inactivated Hippo pathway, overexpression of dephosphorylated YAP1 in nuclei suggests this pathway might be perturbed in C-SCLC which could be a potential mechanism explaining lower response rate to chemotherapy compared with P-SCLC. Further studies are needed to identify the speci c mechanism for dysregulation of HIPPO signaling pathway in C-SCLC.
C-SCLC contains a various kind of non-small cell components with distinct molecular and pathologic characteristics, which might lead to different biological behaviors and treatment vulnerability from P-SCLC. Although C-SCLC is such a highly heterogeneous entity, the treatment choice of C-SCLC is universally consistent with the guidelines for SCLC, usually with a platinum agent and etoposide as common chemotherapeutic regimen. (Moon et al. 2019) Although some NSCLC chemotherapeutic regimens were also tried in C-SCLC, current retrospective data showed no signi cant difference in response rate, progression-free survival (PFS) and OS between patients receiving SCLC and NSCLC chemotherapeutic regimens, but the response rates to both regimen (30% in NSCLC regimen and 38.5% in SCLC regimen) were inferior to that observed in P-SCLC (50%-60% Our study represented rst and largest research to investigate the clinicopathological relevance and prognostic signi cance of YAP1 expression in SCLC strati ed by histological subtypes. For comparison of the prognostic signi cance of YAP1 expression, we applied propensity score matching (PSM) for matching P-SCLC and C-SCLC to balance gender, age, tumor stage and treatment methods since imbalance of these variables may lead to inter-patient bias which might further in uencing prognostic analysis, therefore the reliability of the analytic results increased. Besides, the diagnosis of P-SCLC or C-SCLC was made and con rmed by postoperative pathological section which provided more comprehensive information than cytological or biopsy specimen, therefore increasing the detective rate and accuracy of diagnosis of C-SCLC. Nevertheless, there were some limitation in our study. Firstly, due to the low incidence rate of C-SCLC patients, the number of studied populations was limited resulting a lack of statistical e cacy. Secondly, the assessment of YAP1 expression was based on TMAs which contained limited tissues thus the expression of YAP1 might be higher than current results. Thirdly, since only a small percentage of patients (6.5% in each group) did not receive postoperative adjuvant chemotherapy, we couldn't accurately assess the impact of YAP1 expression on prognosis in this subset of patients. Finally, due to the nature of a respective study, we could not identify the exact correlation of YAP1 expression with treatment response.
In conclusion, we found signi cant higher expression of YAP1 protein in small-cell components of C-SCLC than that of P-SCLC and positive expression of YAP1 was associated worse OS exclusively in C-SCLC. But for P-SCLC, no correlation of YAP1 expression with prognosis was observed. Our ndings indicated the small cell components of P-SCLC and C-SCLC were of different entities with variant YAP1 expression and different prognostic vulnerability which further implicated potential different targetable oncogenic pathway between C-SCLC and P-SCLC. Further comprehensive and basic researches are needed to con rm our results and identify the underlying mechanism by which the YAP1 is overexpressed in C-SCLC and how it in uences the treatment response and prognosis in C-SCLC.

Con icts of interest:
The authors declare that they have no con ict of interest.
Availability of data and material: The dataset analyzed during the current study are available from the corresponding author on reasonable request.
Code availability:   Representative slides of C-SCLC (a, b) and P-SCLC (c, d) stained with hematoxylin and eosin (H&E) (a, c) and immunostained with YAP1 antibody (b, d). The combined SCLC consisted of small cells components (top right corner) combined with adenocarcinoma (a) and was positive for YAP1 in both components (b).
The pure SCLC consisted of pure small cell components (c) and was negative for YAP1 (d).