Prognostic value and characterization of NTRK1 variation by fluorescence in situ hybridization in esophageal squamous cell carcinoma

Abstract

Purpose

Rearrangement of the neurotrophic tyrosine kinase receptor (NTRK) 1 gene is a target of tropomyosin receptor kinase A (TRKA) inhibitors, and its targeted drug (larotrectinib) has been approved by the US Food and Drug Administration. We investigated the existence and prognostic importance of NTRK1 variation in esophageal squamous cell carcinoma (ESCC).

Methods

Fluorescence in situ hybridization of a NTRK1 rearrangement was conducted on 523 ESCC samples through tissue microarrays. Kaplan–Meier curves with log-rank tests were used to evaluate survival.

Results

We identified 8 (1.5%), 35(6.7%) and 109 (20.8%) cases with a NTRK1 rearrangement using 15%, 10% and 5% as cut-off values, respectively. We observed copy number (CN) variation of NTRK1 in some cases: 79 (15.1%) cases had a gain in NTRK1 CN ≥ 3, and 24 (4.6%) cases had NTRK1 CN ≥ 4. A NTRK1 rearrangement at the above-mentioned thresholds was not related to disease-free survival (DFS, P = 0.45, 0.47, 0.87) and overall survival (OS, P = 0.80, 0.74, 0.57), respectively. Gain in NTRK1 CN was associated with a poor prognosis irrespective of whether NTRK1 CN ≥ 4 (DFS, P = 0.015; OS, P = 0.035) or NTRK1 CN ≥ 3 (DFS, P = 0.039; OS, P = 0.025).

Conclusion

A NTRK1 rearrangement occurred rarely in ESCC. The increased CN of NTRK1 might be a prognostic indicator for DFS and OS in patients with ESCC.

Introduction

Esophageal carcinoma (EC) is a common malignant tumor of the digestive tract. EC is the eighth most prevalent malignancy and the sixth leading cause of cancer-related death worldwide (Tong 2018). There are two main histopathologic subtypes of EC: esophageal squamous cell carcinoma (ESCC) and esophageal adenocarcinoma. These two subtypes have different epidemiology and risk factors (Rustgi and El-Serag 2014). EC is a public-health problem in China. It has been estimated that > 250,000 new cases of EC are diagnosed each year, which accounts for half of the cases worldwide (Abnet et al. 2018). ESCC is the predominant subtype in China, comprising ~ 90% of all diagnosed EC (Tong 2018; Abnet et al. 2018). Despite recent advances in various therapeutic approaches, overall survival (OS) at 5 years remains poor for patients with ESCC. Therefore, seeking novel molecular biomarkers for targeted drugs to improve the outcome of patients with ESCC is important.

The neurotrophic tyrosine kinase receptor (NTRK) 1 gene is located on the q-arm of chromosome 1 and encodes tropomyosin receptor kinase A (TRKA), which is a member of the receptor tyrosine kinase (RTK) family of the neurotrophin receptor (Weier et al. 1995). The tropomyosin receptor kinase (TRK) family plays a crucial part in the growth and function of neuronal synapses, memory development/maintenance, and neuronal protection following ischemia or other types of injury (Kaplan and Miller (2000)). Ligands that bind to TRK activate multiple signaling pathways via mitogen-activated protein kinase, phospholipase C-g and phosphatidylinositol 3-kinase to regulate the differentiation and survival of cells (Nakagawara (2001)). Rearrangement involving NTRK1 leads to overexpression of the chimeric protein, resulting in constitutive and ligand-independent activation of its downstream signaling (Vaishnavi et al. 2015).

The NTRK1 rearrangement was first detected in a human colon carcinoma. This discovery was followed by reports in other solid-tumor malignancies, including lung cancer (Vaishnavi et al. 2013), papillary thyroid carcinomas (Greco et al. 2010), soft-tissue sarcomas (Doebele et al. 2015), and malignant melanomas (Lezcano et al. 2018). However, a report on the characteristics of the NTRK1 rearrangement in ESCC has not been reported.

Recently, the US Food and Drug Administration approved the marketing of larotrectinib (Vitkravi®), the first drug to target the NTRK1 rearrangement specifically (Drilon et al. 2018). We hypothesized that the NTRK1 rearrangement may contribute to the proliferation of cancer cells or be involved in critical signaling pathways associated with tumorigenesis in ESCC (as noted in other tumors) and, thus, be a candidate for targeted therapy. ESCC patients might benefit from screening for the NTRK1 rearrangement. We investigated the frequency and the prognostic importance of the NTRK1 rearrangement and variation in gene copy number (CN) in ESCC patients by fluorescence in situ hybridization (FISH).

Materials and methods

Patients and samples

This retrospective study was undertaken on 523 ESCC patients diagnosed and treated at Zhongshan Hospital of Fudan University (Shanghai, China) between 2007 and 2010. Ethical approval was obtained from the Ethics Committee of Zhongshan Hospital of Fudan University (Shanghai, China). Our study was in accordance with the Declaration of Helsinki 1975 and its later amendments. Written informed consent was obtained from patients for use of their surgical specimens for research purposes. Patients were enrolled if they: (i) underwent primary resection (ii); were not undergoing neoadjuvant therapy (neither chemotherapy nor radiotherapy) before resection; and (iii) had complete pathology and clinical records. We excluded patients with very limited tumor tissue and patients with follow-up less than one month. Clinical information (age, sex, smoking history, tumor size, tumor site, and clinical stage) was obtained from review of medical records and pathology reports. Original hematoxylin- and eosin-stained slides were reviewed by two experienced pathologists to obtain information on histology subtype, differentiation, lymph node metastasis, vessels and nerve involvement.

Tissue microarrays (TMAs)

TMAs were constructed as described previously by our research team (Shi et al. 2013). Briefly, regions (2-mm wide and 6-mm long) with a high density of tumor cells from each paraffin-embedded tissue block (“donor tissue rods”) were extracted using a unique TMA sampling tool. Then, the donor tissue rods were planted vertically into the specified position of the recipient block sequentially. Finally, the recipient block was aggregated to align the surface of all planted tissue donor rods through the aggregation instrument. A block of normal esophageal tissue placed adjacent to the tumor from ESCC patients was used to help determine the starting position.

FISH

FISH was carried out on TMA sections of thickness of 4 μm according to manufacturer instructions as described previously (Huang et al. 2018). A dual-color break-apart FISH probe (Empire Genomics, Buffalo, NY, USA) was used to detect a NTRK1 rearrangement. FISH data were evaluated independently by two experienced pathologists blinded to the clinical information of each patient. Slides were scanned under a fluorescence microscope (BX43; Olympus, Tokyo, Japan) equipped with a Microscope Digital Camera (DP73; Olympus). About 100 tumor nuclei from two distinct microscopic areas were evaluated for each patient. Tumor cells were deemed to be positive for the NTRK1 rearrangements if they exhibited a single 3′ signal (green only) or if 5′ and 3′ signals were separated by a distance that was greater than the diameter of one signal. NTRK1 CN was calculated by taking the average number of NTRK1 signals in tumor cells.

Statistical analysis

Statistical analyses were carried out with SPSS v21.0 (IBM, Armonk, NY, USA). Overall survival (OS) encompassed the time from surgery to the date of death due to ESCC. Disease-free survival (DFS) was defined from the date of resection and the date of local, regional, distant recurrence or death. Survival analysis were conducted according to Kaplan–Meier curves. DFS and OS across groups were compared by log-rank tests. Testing of difference between clinicopathologic variables and NTRK1 status was done using the Chi square test or Fisher’s exact test, as appropriate. P < 0.05 (two tailed) was considered significant.

Results

Clinicopathologic characteristics of ESCC patients

We enrolled 523 patients with ESCC. The clinicopathologic characteristics of the study cohort are supplied in Table 1. Patient age ranged from 34 years to 83 (median, 61) years. The cohort comprised 430 (82.2%) males and 93 (17.8%) females. Among them, 321 (61.4%) were non-smokers and 202 (38.6%) had a history of smoking. A total of 248 (47.4%) tumors were located in the middle esophagus, 29 (5.6%) in the upper esophagus, and 246 (47.0%) in the lower esophagus. The histopathologic classification was conducted based on the American Joint Committee on Cancer (AJCC) Staging Manual (eighth edition): 19 (3.6%) were well differentiated, 291 (55.7%) were moderately differentiated, and 213 (40.7%) were poorly differentiated (Fig. 1a). Of all patients, 289 (55.3%) patients had AJCC stage I–II disease and 234 (44.7%) had stage III–IVa disease. Lymph node metastasis was observed in 242 (46.3%) patients. There were 179 (34.2%) and 112 (21.4%) tumors associated with nerve invasion or vascular invasion, respectively (Table 1).

Table 1.

Correlation between clinicopathologic features and NTRK1 status of an ESCC cohort

Characteristics	All patients	NTRK1 rearrangement									NTRK1 copy number (CN)
		≥ 5%			≥ 10%			≥ 15%			CN ≥ 3			CN ≥ 4
	No. (%)	No	Yes	P	No	Yes	P	No	Yes	P	No	Yes	P	No	Yes	P
Sex				0.65			0.72			0.36			0.23			1.00
Female	93 (17.8)	72	21		86	7		93	0		75	18		89	4
Male	430 (82.2)	342	88		402	28		422	8		367	63		410	20
Age				0.37			0.97			1.00			0.46			0.32
< 60	226 (43.2)	183	43		211	15		223	3		188	38		218	8
≥ 60	297 (56.8)	231	66		277	20		292	5		254	43		281	16
Smoking				0.84			0.37			0.27			0.86			0.59
N0	321 (61.4)	255	66		302	19		318	3		272	49		305	16
Yes	202 (38.6)	159	43		186	16		197	5		170	32		194	8
Tumor size				0.48			0.51			1.00			0.88			0.23
I	301 (57.6)	235	66		279	22		296	5		255	46		290	11
II	222 (42.4)	179	43		209	13		219	3		187	35		209	13
Tumor site				0.59			0.77			0.27			0.71			0.44
Upper	29 (5.6)	22	7		28	1		29	0		24	5		29	0
Middle	246 (47.0)	191	55		229	17		244	2		205	41		235	11
Low	248 (47.4)	201	47		231	17		242	6		213	35		235	13
Differentiation				0.23			0.60			0.30			0.32			0.61
Well	19 (3.6)	17	2		17	2		18	1		16	3		18	1
Moderate	291 (55.7)	235	56		274	17		288	3		252	39		280	11
Poor	213 (40.7)	162	51		197	16		209	4		174	39		201	12
Vessel involvement				0.044			0.29			0.68			0.050			0.15
N0	411 (78.6)	333	78		386	25		405	6		345	57		395	16
Yes	112 (21.4)	81	31		102	10		110	2		88	24		104	8
Nerve involvement				0.33			0.027			0.27			0.49			0.59
N0	344 (65.8)	268	76		315	29		337	7		288	56		327	17
Yes	179 (34.2)	146	33		173	6		178	1		154	25		172	7
Lymph node metastasis				0.74			0.092			0.48			0.11			0.10
N0	281 (53.7)	224	57		267	14		278	3		244	37		272	9
Yes	242 (46.3)	190	52		221	21		237	5		198	44		227	15
Clinical stage				0.96			0.24			0.74			0.25			0.073
I–II	289 (55.3)	229	60		273	16		284	5		249	40		280	9
III–Iva	234 (44.7)	185	49		215	19		231	3		193	41		219	15
Disease progression				0.32			0.43			1.00			0.064			0.031
N0	243 (46.5)	197	46		229	14		239	4		213	30		237	6
Yes	280 (53.5)	217	63		259	21		276	4		229	51		262	18
Death of EC				0.71			0.74			0.49			0.048			0.13
N0	253 (48.4)	202	51		237	16		248	5		222	31		245	8
Yes	270 (51.6)	212	58		251	19		267	3		220	50		254	16

Tumor size I tumor length is less than 3 cm; Tumor size II = tumor length is greater than or equal to 3 cm; Smoking no = never smoker

ESCC esophageal squamous cell carcinoma

Fig. 1.

Representative images of H&E-stained (200 × magnification) (a) and FISH of NTRK1 variation in ESCC patients. b Non-NTRK1 rearrangement was shown by a paired red-green fluorescence signal. c A NTRK1 rearrangement was shown as an unpaired red–green fluorescence signal or a single green fluorescence signal. d Gain in NTRK1 CN in tumor cells

NTRK1 rearrangement in ESCC

We identified eight cases (1.5%) from 523 ESCC samples that harbored a NTRK1 rearrangement using 15% as the cut-off value. If 10% and 5% were used as thresholds, the number of cases in which there was a NTRK1 rearrangement was 35 (6.69%) and 109 (20.84%), respectively. Normal paired green and red signals suggested a non-rearranged NTRK1. Separate red and green signals or isolated green signals observed in tumor nuclei (stained blue with 4′,6-diamidino-2-phenylindole) indicated NTRK1 fusion caused by chromosomal rearrangement. Figure 1 illustrates representative patterns of FISH signals of selected normal cases with non-rearranged NTRK1 (Fig. 1b) and cases with a high proportion of NTRK1 rearrangement (Fig. 1c). When using the NTRK1 break-apart probe for FISH detection, a NTRK1 rearrangement was exhibited mainly by a single green fluorescence signal (3′ NTRK1), and the proportion of separate red–green fluorescence signals was low. In addition, a NTRK1 rearrangement was heterogeneous in individual cases. The relationship between a NTRK1 rearrangement and the clinicopathologic features of ESCC is summarized in Table 1. Sex, age, smoking history, tumor size, tumor site, differentiation, lymph node metastasis, clinical stage, disease progression, and death due to ESCC were not correlated statistically with a NTRK1 rearrangement (P > 0.05). However, a NTRK1 rearrangement was associated significantly with vessel involvement (using 5% as a cut-off, P = 0.044) and nerve involvement (using 10% as a cut-off, P = 0.027).

The median duration of follow-up for all patients was 35 (range 2–102) months. DFS and OS at 5 years was 31.9% and 32.7%, respectively. The mean and median time (in months) to DFS were 41.3 and 31.0, and that for OS was 44.6 and 35.0. Kaplan–Meier curves with a log-rank test for DFS and OS were created to evaluate the relationship between a NTRK1 rearrangement and the outcomes of ESCC patients. DFS and OS were not significantly different between the NTRK1-rearranged group and non-NTRK1-rearranged group (Fig. 2). With 5% and 10% as thresholds, the NTRK1-rearranged group had a potentially worse DFS and OS (5%: DFS, P = 0.45; OS, P = 0.80) (10%: DFS, P = 0.47; OS, P = 0.74) than those in the wild-type group (Fig. 2a–d). However, at a threshold of 15%, the NTRK1-rearranged group had a potentially better DFS and OS (DFS: P = 0.87; OS, P = 0.57) than those of the wild-type group (Fig. 2e, f). This inconsistent result may have been due to the low frequency of NTRK1 rearrangement observed.

Fig. 2.

Kaplan–Meier curves of DFS and OS according to a NTRK1 rearrangement in ESCC. With 5% (a, b) and 10% (c, d) as cut-off values, NTRK1-rearranged groups had a potentially worse DFS and OS (5%: DFS, P = 0.45; OS, P = 0.80) (10%: DFS, P = 0.47; OS, P = 0.74) than those in wild-type groups. However, with 15% (e, f) as the cut-off value, the NTRK1-rearranged group had a potentially better DFS (P = 0.87) and OS (P = 0.57) than those in wild-type groups

Variation in NTRK1 CN in ESCC

Among 523 ESCC cases, NTRK1 CN per nucleus ranged from 1.22 to 5.20 (median, 2.31). There were 24 (4.6%) cases with NTRK1 CN ≥ 4, and 81 (15.5%) cases with NTRK1 CN ≥ 3 (Fig. 1d). Table 1 reveals that the variation in NTRK1 CN was not associated significantly with sex, age, smoking, tumor size, tumor site, differentiation, vessel involvement, nerve involvement, lymph-node metastasis, or clinical stage (P > 0.05 for all). However, the variation in NTRK1 CN was associated significantly with disease progression (NTRK1 CN ≥ 4, P = 0.031) and death from ESCC (NTRK1 CN ≥ 3, P = 0.048). In some cases, a NTRK1 rearrangement and variation in NTRK1 CN occurred simultaneously. For 74 patients with a NTRK1 rearrangement ≥ 5% but < 10%, 15 cases had NTRK1 CN ≥ 3 and four cases had NTRK1 CN ≥ 4. For 27 patients with a NTRK1 rearrangement ≥ 10% but < 15%, nine cases had NTRK1 CN ≥ 3 and four cases NTRK1 had CN ≥ 4. For eight patients with a NTRK1 rearrangement ≥ 15%, four cases had NTRK1 CN ≥ 3 and one case had NTRK1 CN ≥ 4 (Fig. 4).

Fig. 4.

Number of cases in which a NTRK1 rearrangement coexisted with copy number variation

The mean and median DFS (in months) for patients with NTRK1 CN ≥ 4 was 27.2 and 22.0, respectively. The mean and median OS (in months) for patients with NTRK1 CN ≥ 4 were 30.8 and 28.0, respectively. The mean and median DFS (in months) for patients with NTRK1 CN ≥ 3 were 32.3 and 24.0, respectively. The mean and median times to OS (in months) for patients with NTRK1 CN ≥ 3 was 36.1 and 32.0, respectively. The impact of the variation in NTRK1 CN on the prognosis of patients was assessed by Kaplan–Meier curves and compared using the log-rank test. With an increase in the CN of NTRK1, the DFS and OS of patients tended to decrease (Fig. 3). Groups with high NTRK1 CN had significantly worse DFS and OS irrespective of whether NTRK1 CN ≥ 4 (DFS, P = 0.015; OS, P = 0.035) or NTRK1 CN ≥ 3 (DFS, P = 0.039; OS, P = 0.025) (Fig. 3c–f). Patients with NTRK1 CN ≥ 4 had marginally worse DFS and OS than those with NTRK1 CN ≥ 3 but ≤ 4 (DFS, P = 0.15; OS, P = 0.30) (Fig. 3a, b).

Fig. 3.

Kaplan–Meier survival curves illustrating the prognostic effects of variation in NTRK1 copy number (CN) in ESCC. a, b With the increase in NTRK1 CN, the DFS (P = 0.072) and OS (P = 0.075) of patients tended to worsen. Patients with increased NTRK1 CN had worse DFS and OS. c, d NTRK1 CN ≥ 3 (DFS, P = 0.039; OS, P = 0.025). e, f NTRK1 CN ≥ 4 (DFS, P = 0.015; OS, P = 0.035)

Discussion

We found that a NTRK1 rearrangement occurred rarely in ESCC, and that these patients may benefit from targeted therapy. Increased NTRK1 CN was associated with worse survival in patients with ESCC. A NTRK1 rearrangement presents in various tumor types. It occurs at a relatively low frequency in common tumors (Cocco et al. 2018), but has a high prevalence in several rare types of cancer and in various pediatric cancers (Gatalica et al. 2019). Increasing evidence suggests that a NTRK1 rearrangement is associated with tumorigenesis, and that patients in which a NTRK1 rearrangement has occurred may benefit from TRKA inhibitors (Khotskaya et al. 2017). Larotrectinib, as the first target drug approved for a NTRK1 rearrangement, has achieved the goal of identical treatment for different diseases (Sidaway 2018). Based on the efficacy of TRKA-targeted inhibitors in various tumors, it can be inferred that ESCC patients with a NTRK1 rearrangement could also benefit from them. Unfortunately, clinical data for EC are lacking. Identification of NTRK1 status in ESCC could provide valuable clues for further clinical trials. We wished to identify the frequency of the NTRK1 rearrangement, and the prognostic importance of NTRK1 variation in ESCC.

The NTRK1 rearrangement was identified through a FISH break-apart probe in 523 ESCC samples. The proportion of ESCC patients with a NTRK1 rearrangement who can benefit from TRKA inhibitors is not known, so we counted cases with different rearrangement ratios separately. Eight (1.5%) cases had a higher proportion (≥ 15%) of a NTRK1 rearrangement, and this observation was consistent with studies looking at other tumor types, such as colorectal carcinoma (0.5%) (Creancier et al. 2015), lung adenocarcinoma (3.3%) (Vaishnavi et al. 2013), and glioblastoma multiforme (1%) (Kim et al. 2014). Pietrantonio et al. (2017) showed that a NTRK1 rearrangement in metastatic colorectal cancer was associated with an unfavorable outcome. Musholt et al. (2019) reported that patients with a NTRK1 rearrangement in sporadic papillary thyroid carcinoma had a worse prognosis than that of patients without this mutation (P = 0.11). Analyses of DFS and OS based on a NTRK1 rearrangement did not reveal a significant prognostic difference. Although a NTRK1 rearrangement has no clear relationship with the prognosis in ESCC, its attractive therapeutic value makes its clinical testing particularly interesting.

Besides NTRK1 rearrangement, we also observed that some cases had polysomy or even presented a small cluster of NTRK1 signals in ESCCs. Mauri and colleagues also demonstrated that a gain in NTRK1 CN was found in other tumors (2018): 2.6% in lung adenocarcinoma, 0.4% in colorectal adenocarcinoma, and 17.6% in biliary tract carcinoma. A gain in gene CN reflects the chromosomal instability of cancer cells and the sustained proliferation and survival of tumor cells is highly dependent upon the activity of the amplified gene (2011). Light et al. (2012) reported that high expression of NTRK1 is closely related to favorable risk factors and outcomes in neuroblastomas patients. In contrast, Dang et al. (2006) found high expression of TRKA in pancreatic cancer to be associated with a poor prognosis. These inconsistent results suggest that high expression of NTRK1 may have different effects on different tumors. Studies on the association between NTRK1 CN and the prognosis for ESCC have not been reported. In our study, the gain in NTRK1 CN was associated with a significantly worse prognosis irrespective of whether NTRK1 CN ≥ 4 (DFS, P = 0.015; OS, P = 0.035) or NTRK1 CN ≥ 3 (DFS, P = 0.039; OS, P = 0.025). Survival analyses of patients with different levels of CNs showed that the prognosis of patients tended to be worse with an increase in NTRK1 CN. Given that the variation in CN of anaplastic lymphoma kinase (ALK), c-ros oncogene 1 receptor tyrosine kinase and epidermal growth factor receptor (EGFR) is a potential therapeutic target or prognostic indicator, exploration of the variation in NTRK1 CN in ESCC is important. Currently, although drugs targeting NTRK1 are directed against the NTRK1 rearrangement, whether patients with increased CN would also benefit from TRKA-targeted inhibitors is not known.

The coexistence of variation in CN and rearrangement of a gene is a frequent event in tumor cells, and CN variation may have an effect on drugs targeting such rearrangement. Algars et al. (2011) found that 25 of 34 patients (73%) with high EGFR CN (≥ 4.0) showed clinical benefit from anti-EGFR therapy, but only four (20%) of 20 patients with low EGFR CN responded to therapy. However, some scholars have suggested that increased gene CN may be a mechanism of acquired drug resistance in targeted therapy. For example, Doebele et al. (2012) found that two crizotinib-resistant patients with an ALK rearrangement had a marked increase in the ALK CN per cell before and after crizotinib treatment. We also observed some cases harboring a NTRK1 rearrangement having a gain in NTRK1 CN. For patients with a NTRK1 rearrangement ≥ 15%, eight cases had NTRK1 CN ≥ 2, four cases had NTRK1 CN ≥ 3, and one case had NTRK1 CN ≥ 4 (Fig. 4). A NTRK1 rearrangement is a predictor of the efficacy of larotrectinib treatment, but the effect of CN variation on rearrangement has not been determined. Future studies will be needed to explore whether variation in the NTRK1 CN has an important impact on the sensitivity to TRKA inhibitors.

In conclusion, this was the first study to: (i) report the characteristics of a NTRK1 rearrangement in ESCC; (ii) identify that NTRK1 CN was associated with the prognosis: patients with increased NTRK1 CN had a worse prognosis. Therefore, NTRK1 variation has value for postoperative management of EC patients.

Abbreviations

NTRK1

Neurotrophic Tyrosine Kinase Receptor 1

TRKA

Tropomyosin receptor kinase A

CN

Copy number

EC

Esophageal carcinoma

ESCC

Esophageal squamous cell carcinoma

RTK

Receptor tyrosine kinase

TRK

Tropomyosin receptor kinase

FISH

Fluorescence in situ hybridization

TMA

Tissue microarrays

DFS

Disease-free survival

OS

Overall survival

ALK

Anaplastic lymphoma kinase

EGFR

Epidermal growth factor receptor

Author contributions

Conceptualization: YH, DJ; Methodology: YH, DJ, QS; Formal analysis and investigation: ZY, HW, JH, XW; Writing—original draft preparation: ZY, HW, ZJ, WC; Writing—review and editing: YH, DJ, JX, JS, HW, LT; Funding acquisition: YH, DJ. All authors read and approved the final manuscript.

Funding

This work was financially supported by Shanghai Natural Science Foundation of China (No. 18ZR1406800), National Natural Science Foundation of China (No. 81702372), Xiamen Science and Technology Project of Fujian Province, China (No. 3502Z20184003), Shanghai Municipal Commission of Science and Technology (No. 19441904000), Shanghai Municipal Key Clinical Specialty (No. shslczdzk01302), and Shanghai Science and Technology Development Fund (No. 19MC1911000).

Data availability

All data generated or analyzed during this study are included in this published article.

Code availability

Not applicable.

Compliance with ethical standard

Conflict of interest

The authors declare that they have no conflict of interest.

Ethics approval

Ethical approval was obtained from the Ethics Committee of Zhongshan Hospital of Fudan University (Shanghai, China). Our study was in accordance with the Declaration of Helsinki 1975 and its later amendments.

Informed consent

Written informed consent was obtained from patients for use of their surgical specimens for research purposes.