Introduction

The longest-known coding gene TTN, which was named CMD1G (cardiomyopathy, dilated 1G), has been reported mutated frequently in many tumor types including lung squamous cell carcinoma, lung adenocarcinoma and colon adenocarcinoma (Kim et al. 2013). However, it is still controversial for TTN as a significant cancer-associated gene only through several bioinformatics analyses. The points of this controversy focus on the false-positive results brought from its giant complicated structure and heterogeneity in the mutational processes (Hofree et al. 2013; Kim et al. 2017; Lawrence et al. 2013). Meanwhile, TTN truncating mutations are identified strongly relating to cardiac and skeletal muscle diseases, but not in cancer, leaving us great conjecture on its roles in cancer (Ceyhan-Birsoy et al. 2013; Herman et al. 2012; Schafer and de Marvao 2017; Watanabe et al. 2016). However, limited studies have been investigated deeply on the relevance among TTN and cancer. One of the limited researches on TTN and cancer has stated that TTN or TP53 mutation in breast cancer does not show strong discriminative power for patient survival. But patients with mutations of certain cancer-associated genes including AKT1, NCOA3, ARID1A or MAP3K1 enjoy a favorable prognosis, and these mutations are exactly mutually exclusive with TTN/TP53 co-mutation (Kim et al. 2017). It suggests that TTN and TP53 double mutation may participate in breast cancer by means of a common downstream pathway with the participation of the other four mentioned genes on the signaling network. Based on these reported studies, the prognostic effects of TTN and TP53 double mutation are explored in this article as well.

NSCLC occupies about 92% of whole-lung cancer cases based on the database utilized in the eighth edition of the TNM classification of lung cancer and takes adenocarcinoma and squamous carcinoma as the majority (Rami-Porta et al. 2014). In addition, TTN and TP53 are the two most common mutant genes in NSCLC including both lung adenocarcinoma and lung squamous cell carcinoma programs queried from The Cancer Genome Atlas (TCGA) lung project (online only). TP53 remains a very well-known high priority in kinds of tumor. On the other hand, combination analyses of TP53 and cancer-associated genes have been commonly conducted to observe clinical outcomes and provide well-founded molecular typing for targeted therapy (Labbe et al. 2017; VanderLaan et al. 2017). However, compared to TP53, controversies around TTN mentioned before, to a certain extent, have discouraged researchers from further analysis. Rare relevance researches of TTN mutation in cancer are reported up to now.

TP53, the well known tumor-suppressor gene, has been detected mutations in the exon region in 30–50% of patients with lung adenocarcinoma and 80% of lung squamous cell carcinoma approximately (Maki-Nevala et al. 2016; Seo et al. 2012). TP53 mutation had been mentioned as a significant independent poor prognostic indicator in NSCLC (Deben et al. 2017; Mitsudomi et al. 1993). Certain research revealed significantly independent poor prognostic effect from “non-disruptive mutation” (HR 6.11 [95% CI 1.43–26.09], P = 0.01), which was defined as lower predicted degree of disturbance to the structure and function of TP53-coding protein, for advanced NSCLC patients (Molina-Vila et al. 2014). This result is incompatible with the reported negative influence of protein changes from TP53 mutation. Another study elucidated a neutral effect of TP53 mutation (exons 5–8) for prognosis of NSCLC patients (HR 0.99 [95% CI 0.77–1.28], P = 0.95) (Ma et al. 2016). As a consequence, the prognostic effect of TP53 in NSCLC remains controversial as well.

This study focuses on the correlation between clinical outcomes of TTN mutations and NSCLC. Attached analyses are explorations of relationships among TTN and TP53 double mutation in NSCLC. This research aims to provide the fundamental basis for further researches on TTN and NSCLC.

Methods

Data collection

Standardized high-quality data including lung adenocarcinoma and lung squamous cell carcinoma programs were collected from The Cancer Genome Atlas (TCGA) online database (http://cancergenome.nih.gov). Both somatic mutation profiles and clinical data were required. Immunostaining image of slides were analyzed for pathological grading of NSCLC samples by two professional investigators who were blinded to the sample information. We artificially defined genetic status as wild-type (WT) or mutant-type (MT) without further subtyping. Lung adenocarcinoma and lung squamous cell carcinoma were, respectively, named LUAD and LUSC in this paper following conventions of TCGA.

Clinical relevance analysis

Clinical relevance of TTN mutation in NSCLC was evaluated with P value, hazard ratio (HR), 95% confidence interval (CI) by univariate analysis and binary logistic regressive model including the following variables: race (Asian, Black or African American, White), histology (squamous, adenocarcinoma), daily smoking amounts, smoking time, gender (male, female), age at diagnosis(< 55, 55–65, 65–75, > 75 years), pathological stage (I, II, III, IV), histological grading [high (well), moderate (moderately), low (poorly)] and history of chemotherapy/radiation/ molecular targeted/ neoadjuvant treatment.

Survival analysis

Overall survival (OS) and disease-free survival (DFS) analysis of patients with TTN mutation or combined mutation were estimated for the assessment of prognostic effect and predictive effect after initial treatment severally. OS, which was the primary end point, was defined as time from random assignment to death from any cause or last follow-up in surviving patients. DFS was defined as time from random assignment to new tumor-associated events after initial treatment or death from any cause or last follow-up in surviving patients. Cox proportional hazards models and Kaplan–Meier methods (Log-rank test) were utilized to conduct survival analysis and draw survival curves. Screening of independent prognostic and predictive factors in mutation types was conducted by a Cox proportional hazards model including frame mutation, stop gain mutation, and missense mutation variates.

Statistical tools

In this study, IBM SPSS software (IBM Corp. Released 2010. IBM SPSS Statistics for Windows, Version 19.0. Armonk, NY: IBM Corp) was applied for univariate and multivariate analysis and Stata Statistical Software (StataCorp.2011.Stata Statistical Software: Release 12.College Station, TX: StataCorp LP) was used for survival analysis and survival curves performing. We set all statistical significance at P < 0.05 and all P values in this study were two sided.

Results

Available patients’ data

1009 cases of NSCLC patients diagnosed with lung adenocarcinoma (n = 517) and lung squamous cell carcinoma (n = 492) from TCGA lung project were available for further analysis. About 989 patients (LUAD n = 504, LUSC n = 485) were eligible for survival analysis. Details of the quantity of available patients relating to different variates are shown, respectively, as LUAD (Online Resource 1) and LUSC (Online Resource 2).

Clinical relevance of TTN in NSCLC

Genetic mutant-type of TTN was detected in 62.0% (626/1009, 95% CI [0.590–0.650]) of total NSCLC samples. About 79.7% (95% CI 0.761–0.832) of lung squamous cell carcinoma and 54.7% (95% CI 0.504–0.590) of lung adenocarcinoma patients harbored TTN mutation. Histology type (multivariable Cox model P < 0.001, HR 3.924 [95% CI 2.164–7.116]) and molecular targeted treatment history (multivariable Cox model P = 0.001, HR 3.102 [95% CI 1.607–5.987]) were proved to be independent influence factors of TTN mutation by a binary logistic regressive model (Table 1). Furthermore, scores on histological grading approached significance (multivariable Cox model P = 0.040). Factors including race, gender, age, pathological stage, daily smoking amounts/time and history of radiation/chemotherapy/neoadjuvant treatment exhibited no independent influence on TTN mutation in NSCLC (Online Resource 4).

Table 1 Binary logistic regression analysis in NSCLC
Full size table

Survival analysis of TTN in LUSC and LUAD

210 out of 485 patients with lung squamous cell carcinoma died and 238 patients had new events after initial treatment including death. Median follow-up times of OS and DFS were 22.2 months (range 0–176 months, mean 32.4 months) and 19.7 months (range 0–176 months, mean 29.9 months) respectively. According to the Cox proportional hazards model (Online Resource 5), TTN mutation was an independent significant favorable prognostic indicator (multivariable Cox model P = 0.013, HR 0.390 [95% CI 0.186–0.817]) (Fig. 1a) but not a predictive indicator (multivariable Cox model P = 0.096) (Fig. 1b) in LUSC. On the other hand, 181 out of 504 LUAD patients died and 237 patients had new events after initial treatment including death. Median follow-up time of OS and DFS was 21.9 months (range 0.1–241.6 months, mean 30.4 months) and 20.2 months (range 0.1–241.6 months, mean 28.5 months) respectively. However, TTN mutation did not manifest significant prognostic (multivariable Cox model P = 0.534) (Fig. 1c) and predictive value (multivariable Cox model P = 0.104) (Fig. 1d) for lung adenocarcinoma.

Fig. 1
figure 1

Unadjusted Kaplan–Meier curves of OS and DFS effect of according to TTN mutation status. a OS analysis in LUSC. b DFS analysis in LUSC. c OS analysis in LUAD. d DFS analysis in LUAD. P value and HR [95% CI] were given by Cox proportional hazards models

Full size image

Prognostic and predictive effect of TTN mutation combined with TP53 mutation

TP53 mutation (including copy number variations) was detected in 86.2% (424/492, 95% CI 0.831–0.892) of LUSC subjects. Around 72.2% of LUSC patients harbored concurrent TTN mutation. Patients with TTN/TP53 double mutation had significantly better OS (multivariable Cox model P = 0.001, HR 1.923 [95% CI 1.330–2.779]) (Fig. 2a) and better DFS after initial treatment (multivariable Cox model P = 0.013, HR 1.565 [95% CI 1.099–2.230]) (Fig. 2b) compared with patients with TP53 single mutation (multivariable Cox model result is shown in Online Resource 6).

Fig. 2
figure 2

Unadjusted Kaplan–Meier curves of OS and DFS effect according to TTN and TP53 mutation status (TTN/TP53 combination) (MT, +; WT, −). a OS analysis in LUSC. b DFS analysis in LUSC. c OS analysis in LUAD. d DFS analysis in LUAD

Full size image

TP53 mutation was found in 51.1% (264/517, 95% CI 0.467–0.554) of LUAD patients meanwhile 35.6% of a total 517 patients had TTN/TP53 double mutation. However, there was no significant difference in OS (multivariable Cox model P = 0.801) (Fig. 2c) and DFS (multivariable Cox model P = 0.666) (Fig. 2d) on the basis of mutation status (multivariable Cox model result is shown in Online Resource 7).

Mutation types of TTN in lung squamous cell carcinoma

In 392 mutation subjects out of 492 lung squamous cell carcinoma cases, one subject could have a different number of TTN single-nucleotide variations from the other (range 1–12 loci, individual case 17 or 30 loci) and mutation types of TTN in one patient could be various. Among the 392 cases, 104 cases had only one nucleotide-site variation of TTN while 288 of subjects which constituted a majority of mutation cases observed TTN nucleotide variations in more than 1 locus (Fig. 3a). Kaplan–Meier survival analysis showed that patients with variations of two or more loci obtained only a better overall survival benefit than patients with TTN wild-type and only one site alteration [Log-rank OS P = 0.0160 (Fig. 3b), DFS P = 0.1777 (Fig. 3c)].

Fig. 3
figure 3

Aggregate analysis of TTN mutation types in lung squamous cell carcinoma. a Pie chart (n = 492) of patients with more than one TTN nucleotide variation locus and other status. b, c Unadjusted Kaplan–Meier curves of OS/DFS effect according to amounts of TTN nucleotide variations. d Pie chart (n = 1232) of TTN mutation types. e, f Unadjusted Kaplan–Meier curves of OS/DFS effect according to mutation types

Full size image

A total of 1232 single-nucleotide variations were found in above 392 patients. According to the transcription consequence, missense mutations accounted for the top most of mutation types (n = 761). The second ranked type was synonymous alterations (n = 265), followed by intron alterations (n = 105), stop gain mutations (n = 53), and frameshift mutations (n = 32). Besides, a small proportion of patients were found as in-frame mutations (n = 4), 3′ Prime UTR alterations (n = 2) and splice site mutations (n = 10) (Fig. 3d). For patients with mutations of about moderate-to-high degree of disturbance to the structure and function of TTN-coding protein predicted by TCGA (online only), 314 patients out of 392 harbored missense mutations while 43 with stop gain mutations and 25 with frame mutations. Moreover, for patients with mutations of low degree of disturbance, 159 patients bore synonymous alterations and 92 harbored intron alterations. The other mutations including splice site mutations and 3′ Prime UTR alterations were not discussed alone because of low incidence (Table 2). Multivariate Cox regression model identified that missense mutation was an independent favorable prognostic indicator for lung squamous cell carcinoma (P = 0.025, HR 0.724 [95% CI 0.546–0.960]) (Fig. 3e) but not a significant independent predictive indicator (P = 0.068, HR 0.781 [95% CI 0.599–1.018]) (Fig. 3f).

Table 2 COX analysis of TTN mutation types in lung squamous cell carcinoma
Full size table

Prognostic and predictive effect of TTN/EGFR, TTN/ALK, TTN/ROS1, TTN/RET combination in LUSC

EGFR, ALK, ROS1, and RET mutations (including Copy number variations) were detected, respectively, in 18.5% (91/492, 95% CI 0.151–0.219), 12.6% (62/492, 95% CI 0.097–0.155), 15.4% (76/492, 95% CI 0.122–0.187) and 14.0% (69/492, 95% CI 0.109–0.171) of LUSC subjects (Online Resource 8). As shown in Fig. 4 and Online Resource 9–10, patients with TTN/EGFR (P = 0.011, HR 0.526 [95% CI 0.320–0.863]), TTN/ROS1 (P = 0.017, HR 0.551 [95% CI 0.338–0.900]), TTN/RET (P = 0.035, HR 0.587 [95% CI 0.358–0.962]) double-mutant tumors had significantly better OS compared with patients with double WT tumors. Meanwhile, patients with TTN/RET double-mutant tumors had significantly better DFS (P = 0.022, HR 0.572 [95% CI 0.354–0.923]) compared with patients with double WT tumors. Furthermore, patients with TTN single mutation obtained significantly better OS and DFS benefit compared with patients with double WT tumors throughout the whole TTN/oncogenes combination analysis (most P < 0.005). The Cox model adjusted by the four oncogenes suggested that TTN remained an independent prognostic factor for lung squamous cell carcinoma (P = 0.022, HR 0.572 [95% CI 0.354–0.923]) (Online Resource 11).

Fig. 4
figure 4

Unadjusted Kaplan–Meier curves of OS and DFS effect of according to TTN and oncogenes mutation status in lung squamous cell carcinoma (MT, +; WT, −). a OS analysis of TTN/EGFR combination. b DFS analysis of TTN/EGFR combination. c OS analysis of TTN/ALK combination. d DFS analysis of TTN/ALK combination. e OS analysis of TTN/ROS1 combination. f DFS analysis of TTN/ROS1 combination. g OS analysis of TTN/RET combination. h DFS analysis of TTN/RET combination

Full size image

Discussion

In this study, subsequent analyses on lung squamous cell carcinoma and lung adenocarcinoma are conducted separately, since TTN mutation shows significant difference between lung squamous cell carcinoma and lung adenocarcinoma. Although TTN mutation also shows significant relevance with gender, smoking situations through univariate analysis (Online Resource 3), the binary logistic regressive model does not identify them as independent influence factors for TTN mutation. Meanwhile, patients with TTN mutation are prone to possessing a history of molecular targeted treatment which suggests that TTN mutation may tend to co-mutate with oncogenes. However, the details about molecular targeted drugs are unknown and we further discuss the combination analysis of TTN and common oncogenes later. Besides, TTN mutation is significantly different between poorly differentiated tumor and moderately/well differentiated tumor (P = 0.040) which cannot explain why TTN plays a role in better survival. More messages are required to explain the relationship between TTN and significant clinical findings.

Based on these results, we further focus on the survival analysis of TTN in the two histological subtypes of NSCLC, respectively. As far as our information goes, it is the first time to assess prognostic and predictive value of TTN single and dual mutation in NSCLC. TTN manifests no prognostic and predictive effect for lung adenocarcinoma. However, this situation is completely opposite in lung squamous cell carcinoma. We firstly report that TTN mutation displays favorable prognostic value in lung squamous cell carcinoma (P = 0.013), but does not reach the conventional significance level of predictive value (P = 0.096). Meanwhile, patients with TTN/TP53 double mutation would have both better OS and DFS benefit compared with patients with TTNWT/TP53MT status which means that TTN and TP53 mutation may have a synergistic effect in lung squamous cell carcinoma. Our work suggests that TTN mutation influences the prognosis of patients with lung squamous cell carcinoma and mutation means beneficial.

It is not the first time for researchers to find gene mutation indicating good prognosis. Previous study has mentioned that patients with PIK3CA mutation enjoy a significantly better overall survival benefit (P = 0.042) than non-mutated cases in lung squamous cell carcinoma (McGowan et al. 2017). Another research has showed that patients with colorectal cancers would obtain a better prognosis when they harbored POLE proofreading domain mutation (HR 0.34, P = 0.006) which was possibly caused by increasing expression of cytotoxic T-cell markers and effector cytokines or enhancing CD8+ lymphocyte infiltrating (Domingo et al. 2016). Besides, NSCLC patients with lymphatic spread (P = 0.014) or advanced NSCLC patients (P = 0.011) would possess favorable prognosis when they harbor A type 3 expression (Schmidt et al. 2016).

Then we discuss the mutation types of TTN in lung squamous cell carcinoma to explain this phenomenon. It is difficult to point out which site variation happened most since the same nucleotide locus variation could happen in two samples at most. Single sample could have various mutation types at the same time. Meanwhile, according to corresponding transcription ending, missense mutations and synonymous variations are common. It is interesting to find that patients with two or more TTN nucleotide variations possess better prognosis compared to those patients with only one site variation and wild-type (P = 0.016). Besides, missense mutation is an independent prognostic indicator for good prognosis of lung squamous cell carcinoma (P = 0.025, HR 0.72) rather than a predictive indicator (P = 0.068, HR 0.781) which is exactly the same as the previous conclusion. Considering missense mutations occurring in 85% of mutated cases, it strongly indicates that two or more TTN nucleotide variations indicating better prognosis may owe to missense mutations. It is well known that a single-nucleotide missense change commonly results in a codon which codes for a different amino acid (Minde et al. 2011). Thus, whether mutant peptide segments somehow triggering certain immune reaction or improving some response to drugs to help patients obtain better survival remains unknown. Further researches are required to confirm this conjecture.

Finally, we talk about the co-mutation of TTN with well known oncogenes including EGFR, ALK, ROS1, and RET. The results show that TTN appears to play a central role in these co-mutation effects. The Cox model confirms that TTN mutation is still an independent prognostic factor for lung squamous cell carcinoma after adjustment by above oncogenes. We also list the co-mutation results (single-nucleotide variation only) of the comparison of these four oncogenes. Coexistence of possible pathogenic single-nucleotide variations rarely occurs between EGFR, ALK, ROS1, and RET.

In summary, TTN missense mutation acts a beneficial prognostic role in lung squamous cell carcinoma. This phenomenon is not suitable for lung adenocarcinoma. TTN/TP53 double mutation has favorable prognostic and predictive value for lung squamous cell carcinoma as well. Through the popularization and development of sequencing technology and other technology, further clinical trials and basic researches will be required to explain the relevance and mechanism of TTN and lung squamous cell carcinoma.