The clinicopathological and prognostic significance of mTOR and p-mTOR expression in patients with non-small cell lung cancer: A meta-analysis

Background:

The mammalian target of rapamycin (mTOR) has a crucial role in carcinogenesis, angiogenesis, cellular proliferation, and metastasis; however, its significance in non-small cell lung cancer (NSCLC) remains contentious. Consequently, this study aims to assess the clinicopathological and prognostic importance of mTOR/p-mTOR expression in NSCLC.

Methods:

Literature retrieval was undertaken by searching English databases PubMed, EMBASE, Web of Science, and Cochrane Library as well as Chinese databases CNKI, Wan Fang, and VIP for full-text publications that satisfied our eligibility criteria up to November 2021. STATA 12.0 was used to conduct statistical analysis (STATA Corporation, College Station, TX).

Results:

This meta-analysis includes a total of 4683 patients from 28 primary publications. mTOR/p-mTOR expression was associated with sex (OR = 0.608, 95% CI: 0.442–0.836), lymph node metastasis (OR = 2.084, 95% CI: 1.437–3.182), and CEA (OR = 1.584, 95% CI: 1.135–2.209), but not with age, histological type, depth of tumor invasion, distant metastasis, TNM stage, differentiation degree, tumor size, or smoking. In addition, the expression of mTOR/p-mTOR is related to shorter overall survival in NSCLC patients (HR = 1.415, 95% CI: 1.051–1.905).

Conclusion:

Positive mTOR/p-mTOR expression was substantially correlated with unfavorable conditions on the sex, lymph node metastases, and CEA levels. mTOR/p-mTOR may indicate a bad prognosis for NSCLC. The current findings must be confirmed and changed by other high-quality research employing a multivariate analysis on bigger sample size.

1. Introduction

Non-small cell lung cancer (NSCLC) represents the most frequent type of respiratory tumor in clinical practice, with a percentage of 85% of all primary lung carcinomas.^[1] Epidemiological data show^[2] that by 2022, lung cancer will remain the leading cause of cancer-related death in China and emerge as the malignancy with the highest morbidity and mortality. In the past decade, low-dose CT-based lung cancer screening has significantly reduced the mortality of patients,^[3] and the introduction of molecularly targeted drugs and immune checkpoint inhibitors has improved the prognosis of patients with advanced lung cancer,^[4] but drug resistance and immune-related adverse effects have limited their clinical application.^[5,6] So it is crucial to identify novel targets to enhance the prognosis of patients with non-small cell lung cancer.

The mammalian target of rapamycin (mTOR) is a highly conserved serine/threonine (Ser/Thr) kinase that serves as a critical regulatory protein in proper cell growth, survival, metabolism, development, and angiogenesis.^[7] mTOR regulates protein production via regulating transcription, translation, and ribosome biosynthesis, a process that affects cell growth and size and is essential for cell division.^[8] Therefore, excessive activation of mTOR will result in elevated levels of cellular metabolism, sustained cell growth and proliferation, and prolonged cellular longevity, hence promoting the development of cancer and metabolic disorders.^[9] Inhibition of mTOR can arrest the cell cycle in the G1 phase, as well as block mTOR-related signaling pathways, resulting in anti-proliferation, anti-inflammatory, autophagy-inducing, and apoptosis-inducing effects.^[10] This provides a crucial direction for the development of anti-cancer and immunosuppressive drugs. In recent years, mTOR-targeted medicines have been widely developed and employed in preclinical and clinical trials for cancer.^[11–14] As the first allosteric inhibitor of mTOR, Sirolimus, for instance, has been employed as an anti-tumor, anti-cardiovascular reconstruction, and anti-coronary restenosis therapy.^[15,16] The discovery of mTOR inhibitors for use in cancer treatment has also demonstrated remarkable effectiveness. Tesirolimus is now licensed to treat advanced renal cell carcinoma, mantle cell lymphoma, platinum-refractory/drug-resistant ovarian cancer, and advanced/recurrent endometrial cancer.^[17] Everolimus has shown partial effectiveness in the treatment of subependymal giant cell astrocytoma,^[18] renal angiomyolipoma,^[19] advanced nonfunctional pulmonary/gastrointestinal neuroendocrine tumor,^[20] and advanced NSCLC.^[21] In addition, the use of mTOR inhibitors in neuroblastoma, pediatric solid tumors, and sarcomas is still in the clinical trials phase.^[22,23] In conclusion, mTOR, as the master switch of cellular metabolism, is an efficient anticancer therapeutic target. In cancers, blocking overactive mTOR signaling and associated pathways can cause potent antitumor effects. Currently, the therapeutic advantages of targeted cancer therapy are restricted to a minority of patients whose tumors are often caused by particular genetic abnormalities inside tumor cells.^[24] Therefore, targeting the mTOR signaling pathway to block translation start is emerging as a possible treatment strategy. Results indicate the possible applicability of mTOR-targeted medicines as additional new chemotherapy components for a certain subgroup of lung cancer patients. Therefore, this meta-analysis is based on the theoretical basis that the role of mTOR in autophagy has an important impact on tumor occurrence and development, and further explores the correlation between mTOR expression and cancer progression and prognosis in clinical studies, which provides a practical basis for the development and research of mTOR targeted inhibitors.

Among the intracellular pathways that are abnormally elevated in cancer is the phosphoinositide 3-kinase (PI3K)-Akt-mTOR pathway. Oncogene-activating mutations, oncogene amplification, or tumor suppressor gene inactivation have been found in many malignancies, resulting in dysregulation of the PI3K-Akt-mTOR signaling pathway.^[25] Data from current studies show that this pathway is activated in approximately 70% of ovarian or breast cancer^[26]; abnormally activated this pathway is seen in 90% of lung adenocarcinomas (ADC) and 40% of lung squamous cell carcinomas^[27]; 63.5% of hepatocellular carcinoma reported immunoreactivity of phosphorylated mTOR on tissue sections, and a significant association was found between p-mTOR expression and tumor size, metastasis and prognosis.^[28] In addition, phosphorylation of PDK1 (phosphoinositide-dependent kinase-1) at Thr308 and mTORC2 (mTOR complex 2) at Ser473 was observed to be associated with worse overall survival in patients with AML (acute myeloid leukemia).^[29,30] Numerous clinical studies have investigated the role and potential prognostic value of mTOR and p-mTOR in a variety of cancers. In a meta-analysis that pooled data from 915 patients with esophageal squamous cell carcinoma, mTOR/p-mTOR expression was found to be significantly associated with worse overall survival, disease-free survival, and cancer-specific survival.^[31] In addition, breast cancer patients with mTOR overexpression had a threefold increased risk of recurrence compared to those without mTOR overexpression.^[32] Meanwhile, studies on the relationship between mTOR/p-mTOR expression and gastric cancer, breast cancer, colorectal cancer, nasopharyngeal carcinoma, and urinary system tumors have shown that mTOR or p-mTOR may be considered promising markers for predicting aggressiveness and prognosis of cancer.^[33–37] In NSCLC, dysregulation of Akt/mTOR signaling pathway activity is one of the primary factors of carcinogenesis. mTOR promotes tumorigenesis and progression through activation of the downstream eIF4 complexes.^[38] As confirmed by immunohistochemical (IHC) staining, p-mTOR activation was observed in 50% to 60% of lung cancer patients, including 66% of NSCLC.^[39] Among them, up to 90% of ADC, 60% of large cell carcinoma, and 40% of squamous cell carcinoma (SCC) were positive for p-mTOR, suggesting that mTOR may have a role in the morphogenesis or differentiation of glandular structure.^[40] In addition, many studies have demonstrated that mTOR may be a factor in lymph node metastasis in SCC.^[39] Collectively, these findings indicate that the stimulation of mTOR-mediated signaling may be clinically aggressive in specific populations. Therefore, clarifying the association between the clinicopathological features of various malignancies and mTOR, evaluating the sensitivity of each clinical lung cancer case to mTOR-targeted pharmacological therapy, and adopting specialized therapy for distinct subtypes of tumors may be a more valuable strategy.

Numerous research has been conducted on the association between the activation of mTOR and the clinicopathological characteristics and prognosis of NSCLC, but no clear result has been reached. In lung squamous cell carcinoma, mTOR/p-mTOR phosphorylation was positively connected with lymph node metastasis and distant metastasis, according to one research.^[39,40] But it is unclear why lymph node metastasis is related to mTOR activation. S6K and 4E-BP1 are downstream signaling molecules of mTOR, and mTOR/S6K activation is associated with the differentiation, maintenance, and potential morphogenesis of some fractions of lung cancer (especially ADC).^[41] An increase in eIF4F formation can boost translation initiation and cell proliferation.^[38] The activation of S6K and 4E-BP1 has been found to determine cisplatin resistance in NSCLC. Nonetheless, 4E-BP1 or S6K activity has little prognostic value for lymph node metastasis or overall survival in NSCLC, whereas the influence of mTOR on the prognosis of NSCLC is still under investigation.^[39] Numerous clinical papers have studied the involvement of mTOR and p-mTOR in NSCLC; however, due to the variability of IHC standards, the conclusions are inconsistent, and several contentious findings have not been adequately addressed. The predictive usefulness of the mTOR/p-mTOR expression and its association with certain common clinicopathological characteristics of NSCLC are still debatable. To assess the potential of mTOR/p-mTOR as a biomarker relevant for NSCLC progression and prognosis, to describe the differences in clinicopathological features and prognosis of mTOR/p-mTOR overexpression based on country, race, and specific populations, and to investigate its clinical role, we conducted the current meta-analysis, synthesizing appropriate evidence from homogenous studies to draw general conclusions.

2. Materials and methods

2.1. Protocol

This work was registered with PROSPERO (registration number: CRD42021292457) and followed the latest PRISMA 2020 statement (Preferred Reporting Items for Systematic Reviews and Meta-Analysis).^[42]

2.2. Inclusion criteria

Articles were considered eligible if they met all the following standards: Patients were pathologically diagnosed with primary non-small cell lung cancer, without other malignant tumors, without any preoperative treatment such as chemotherapy/radiation, and with no age or gender restrictions; The expression of mTOR/p-mTOR should be analyzed independently rather than in collaboration with other biomarkers. IHC was the only experimental method used to stain NSCLC specimens; The study has sufficient data or adequate information to assess the odd ratios (ORs) to estimate the correlation between mTOR/p-mTOR expression and clinicopathological characteristics of NSCLC, or to assess the hazard ratios (HRs) and 95% confidence intervals (CIs) of overall survival (OS); The included studies are cohort study or case-control study.

2.3. Exclusion criteria

The following articles should be excluded: Animal experiments, case reports, reviews, conference abstracts, letters, and repeated data; Positive expression of mTOR/p-mTOR was not stained by IHC; Insufficient information to estimate the prognostic and clinicopathological correlation; Not written in English or Chinese.

2.4. Literature search strategy

We conducted a thorough literature search for relevant papers published through November 2021 in the following databases: PubMed, Web of Science, Cochrane Library, Embase, CNKI (China National Knowledge Infrastructure), WanFang, and VIP. No language or publication date restrictions were imposed on the search. In reference to a comparable meta-analysis examining the clinicopathological features and prognostic usefulness of biomarkers,^[43,44] the following search phrases were employed: lung cancer or lung tumor or bronchogenic carcinoma or non-small cell lung cancer; mammalian target of rapamycin or mTOR or p-mTOR or phosphorylated mammalian target of rapamycin; prognosis or prognostic or survival. In addition, the references of relevant literature were manually examined to verify that the literature search was exhaustive.

2.5. Literature retrieval and quality assessment

Two researchers evaluated the listed studies independently using the Newcastle–Ottawa Scale (NOS).^[45] The scale had eight items divided into three sections, with a total of nine points. Included were studies with or exceeding a score of 6. Discrepancies in appeal results were discussed or consulted with a third investigator.

2.6. Data extraction

All data were extracted and cross-checked independently by two reviewers from eligible studies with a predefined table. If there were inconsistencies that affected the quality of pooled results, a third investigator assessed and determined the data independently. These following variables have been retrieved: Publication information including first author, publication year, country, and language; Clinical information including total sample size, the number of patients with positive and negative expression of mTOR/p-mTOR, genders, ages, pathological type, clinical stage, pathological grade, lymph node metastasis, tumor size, differentiation, smoking status, CEA level; prognostic indicators including Endpoint events, follow-up periods, HRs and 95% CI (including univariate and multivariate analysis); Experimental data including experimental materials, detecting methods, IHC techniques (antibodies types, companies, and dilution), cutoff values.

2.7. Statistical analysis

Based on the level of mTOR/p-mTOR detection, the patients were divided into high-expression and low-expression groups. Pooled OR and 95% CI were adopted to evaluate the correlation between mTOR/p-mTOR expression and clinicopathological characteristics. Pooled HR and 95% CI were calculated as the effective index to estimate the impact of mTOR/p-mTOR expression on the OS/PFS of the patients.^[46] If HR > 1 and its 95% CI did not overlap with 1, this indicated that positive mTOR/p-mTOR expression was a risk factor for the prognosis of NSCLC, namely, the higher the mTOR/p-mTOR expression, the worse the prognosis of patients; if HR < 1 and 95% CI did not overlap with 1, the opposite was true. The heterogeneity was measured by the Q test and the inconsistency index I². If the P value was less than 0.1 or I² was greater than 50%, the heterogeneity was considered significant, and the random-effect model was adopted; otherwise, if the heterogeneity was not significant, the fixed-effect model was adopted.^[47] To analyze the reliability of the pooled result, sensitivity analysis was performed by omitting every single study.^[48] Additionally, subgroup analysis was also conducted; Begg’s funnel plot and Egger’s s bias indicator test were used to assessing the publication bias.^[49] All P values were two-tailed in this meta-analysis, and P < .05 was considered statistically significant. Data were processed using Stata version 12.0 (Stata Corporation, College Station, TX).

3. Results

3.1. Literature search

1263 pieces of relevant literature were retrieved; 178 duplicate articles were removed; 119 articles were checked automatically using Endnote software and 59 articles were checked manually; By reading the title and abstract, 1036 articles, including reviews, case reports, animal experiments, and unrelated articles, were removed; By carefully reading the full text, 21 articles were eliminated, including 1 article not written in English or Chinese, 3 reviews and conference abstracts, 7 repeated data studies, 8 studies that did not provide mTOR/p-mTOR positive expression rate or prognosis-related data, and 2 studies that did not directly report HR value and its 95% CI but only gave K-M survival curve,^[50,51] and a total of 28 studies were determined to be included.^[52–79] Figure 1 shows the flowchart of the literature search.

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram for literature retrieval.

3.2. Study characteristics

This meta-analysis includes a total of 28 papers, 13 of which were authored in English and 15 in Chinese. The 28 included studies were published between 2008 and 2021, contained a total of 4683 patients, and had sample sizes ranging from 35 to 574. 23 of the 28 papers examined the association between mTOR/p-mTOR expression and clinicopathological characteristics of NSCLC,^{[52–57,61–77]} whereas 10 examined the relationship between mTOR/p-mTOR expression and prognosis.^{[53,54,58–60,63,76–79]} The positive expression rate of mTOR/p-mTOR was measured by immunohistochemical staining of paraffin-embedded tissues (IHC). Tables 1–3 outlines the general features and experimental materials of the included research.

Table 1.

The general characteristics of studies on mTOR expression.

First author (year)	Country	CP features/Prognosis	No. of patients (mTOR PE/NE)	Genders (male/female)	Age	Histological type (SCC/ ADC/Other)	Stage (Ⅰ+ⅠⅠ/ⅠⅠⅠ+Ⅳ)
Dan Liu 2011	China	Y/Y	134 (106/28)	138/34	<60 (81)/≥60 (91)	77/75/20	85/87
Oh Mee-Hye 2012	Koreans	Y/N	574 (376/198)	398/176	<65 (282)/≥65 (292)	297/227/50	417/157
Wang L 2012	China	Y/N	78 (51/27)	66/12	≤60 (37)/> 60 (41)	34/44	NA
Kim HS 2012	NA	Y/N	245 (161/84)	194/51	<64 (130)/≥64 (115)	91/154	167/78
Valsamo K 2009	USA	N/Y	167 (94/73)	NA	64	97/0	NA
Dhillon T 2010	Italy	N/Y	134 (73/61)	117/17	<65 (77)/≥65 (57)	41/56/37	NA
Gately K 2012	UK	N/Y	141 (101/40)	85/56	<65 (67)/≥65 (74)	60/67/14	141/0
Guo K 2014	China	Y/N	82 (44/38)	46/36	<60 (58)/≥60 (24)	43/39	NA
Xu YY 2021	China	Y/N	35 (27/8)	12/23	<65 (23)/≥65 (12)	33/2	NA
Chen ZH 2013	China	Y/N	42 (21/21)	22/20	≤60 (15)/ >60 (27)	23/19	34/8
Guo ZQ 2018	China	Y/N	105 (79/26)	65/40	≤60 (32)/ >60 (73)	35/70	63/42
Hu ST 2010	China	Y/N	80 (42/38)	64/16	<60 (35)/≥60 (45)	0/80	67/13
Huang W 2019	China	Y/N	246 (136/110)	150/96	≤60 (123)/>60 (123)	154/92	NA
Wang XC 2020	China	Y/N	93 (43/50)	89/4	≤65 (63)/>65 (30)	0/93	NA
Yuan C 2017	China	Y/N	80 (46/34)	59/21	<60 (33)/≥60 (47)	32/31/17	36/44
Zhang JY 2020	China	Y/N	80 (49/31)	56/24	NA	NA	NA
Cai WK 2014	China	N/Y	138 (101/37)	83/55	<65 (63)/≥65 (75)	68/59/11	NA
First author (year)	Node metastasis (Y/N)	Differentiation (G1 + G2/G3)	Estimates	Antibody	Company	Dilution	Cut off value
Dan Liu 2011	NA	100/72	OR, HR	RMA	CST, Beverly, MA	01:50	≥2
Oh Mee-Hye 2012	240/334	NA	OR	RMA	Danvers, MA	NA	≥3
Wang L 2012	23/55	57/21	OR	RMA	Beverly, MA	01:50	>2
Kim HS 2012	106/139	206/39	OR	RMA	Burlingame, Calif.	1:100	>2
Valsamo K 2009	NA	NA	HR	RMA	CST	1:50	>28
Dhillon T 2010	34/100	NA	HR	RMA	CST, Danvers, MA	01:50	>30
Gately K 2012	NA	NA	HR	RMA	CST, Danvers, MA	01:50	>30
Guo K 2014	38/44	56/26	OR	MMA	Abcam	NA	>2
Xu YY 2021	NA	NA	OR	RMA	Bioss, Beijing	1:100	≥1
Chen ZH 2013	20/22	NA	OR	RMA	Bioss, Beijing	NA	≥3
Guo ZQ 2018	62/43	85/20	OR	RMA	Boster, Wuhan	01:50	>3
Hu ST 2010	65/15	68/12	OR	RMA	CST	1:100	>1
Huang W 2019	67/179	NA	OR	RMA	NA	NA	≥2
Wang XC 2020	33/60	NA	OR	NA	America SC	NA	>2
Yuan C 2017	33/47	56/24	OR	RMA	Abcam	NA	≥1
Zhang JY 2020	NA	NA	OR	RMA	Shanghai YSRIBIO	NA	NA
Cai WK 2014	91/47	89/49	HR	NA	NA	NA	NA

ADC = adenocarcinoma, CP = clinicopathological, HR = hazard ratio, MMA = mouse monoclonal antibody, NA = not available, NE = negative expression, OR = odd ratio, PE = positive expression, RMA = rabbit monoclonal antibody, SCC = squamous cell carcinoma.

Table 3.

Prognostic characteristics of studies on mTOR/p-mTOR expression.

First author (year)	Analysis	Univariate HR (95% CI)	Multivariate HR (95% CI)	Endpoints
mTOR
Dan Liu 2011	U	0.645 (0.377–1.103)	NA	OS
Valsamo K 2009	M	NA	0.48 (0.24–0.98)	OS
Dhillon T 2010	U&M	1.77 (1.17–2.73)	1.66 (1.01–2.74)	OS
Gately K 2012	U&M	1.85 (0.98–3.49)	2.18 (1.12–4.23)	OS
Cai WK 2014	M	NA	3.697 (1.687–5.195)	OS
p-mTOR
Dan Liu 2011	U&M	1.917 (1.349–2.724)	3.299 (1.928–5.645)	OS
Zhang Y 2013	M	NA	2.642 (1.157–4.904)	OS
Liu HB 2008	M	NA	0.686 (0.274–1.721)	OS
Yoshizawa A 2010	U	1.01 (0.75–1.36)	NA	OS
Shimizu K	U	1.079 (0.597–1.948)	NA	OS, PFS
Shimizu K	U	1.475 (0.868–2.505)	NA	OS, PFS
Gold KA	M	NA	0.662 (0.460–0.952)	OS

CI = confidence interval, HR = hazard ratio, M = multivariate analysis, NA = not available, OS = overall survival, PFS = progression-free survival, U = univariate analysis.

Table 2.

The general characteristics of studies on p-mTOR expression.

First author (year)	Country	CP features/prognosis	No. of patients (p-mTOR PE/NE)	Genders (male/female)	Age	Histological type (SCC/ADC/Other)
Dan Liu 2011	China	Y/Y	172 (89/83)	138/34	<60 (81)/≥60 (91)	77/75/20
Zhang Y 2013	China	Y/Y	120 (46/74)	81/39	<60 (54)/≥60 (66)	57/63
Lu J 2020	China	Y/N	341 (194/147)	256/85	<55 (137)/≥55 (204)	182/159
Shin E 2015	USA	Y/N	408 (180/228)	289/119	NA	250/158
Liu HB 2008	China	Y/Y	59 (24/35)	43/16	<60 (31)/≥60 (28)	28/27/4
Xu HR 2018	China	Y/N	82 (54/28)	62/20	≤60 (36)/>60 (46)	34/48
A Wang DZ 2016	China	Y/N	49 (37/12)	36/13	<60 (30)/≥60 (19)	20/29
Huang W 2019	China	Y/N	246 (111/135)	150/96	≤60 (123)/>60 (123)	154/92
Wang XC 2020	China	Y/N	93 (47/46)	89/4	≤65 (63)/>65 (30)	0/93
Xu HR 2018	China	Y/N	82 (56/26)	64/18	≤60 (44)/>60 (38)	40/42
Zhang XY 2019	China	Y/N	96 (64/32)	62/34	≤60 (50)/>60 (46)	46/50
Yoshizawa A 2010	USA	Y/Y	276 (129/147)	NA	NA	138/138
Shimizu K 2014	Japan	Y/Y	104 (41/63)	64/40	≤65 (31)/>65 (73)	66/38
Shimizu K 2014	Japan	N/Y	204 (56/148)	119/85	≤65 (67)/>65 (137)	142/62
Gold KA 2014	USA	N/Y	370	186/184	NA	227/126/17
First author (year)	TNM Stage (Ⅰ+ⅠⅠ/ⅠⅠⅠ+Ⅳ)	Node metastasis (Y/N)	Estimates	Antibody	Company	Dilution	Cut off value
Dan Liu 2011	85/87	NA	OR, HR	ser2448, RMA	CST, Beverly, MA	1:100	≥2
Zhang Y 2013	72/48	69/51	OR, HR	NA	CST	01:50	≥2
Lu J 2020	168/173	202/139	OR	IgG isotype-matched antibody	NA	1:100	≥2
Shin E 2015	NA	25/383	OR	NA	NA	NA	≥4
Liu HB 2008	41/18	38/21	OR, HR	ser2448, RMA	CST	1:100	≥2
Xu HR 2018	NA	26/56	OR	RMA	Santa Cruz	NA	>3
A Wang DZ 2016	36/13	22/27	OR	RMA	Shanghai Yuanmu	NA	≥2
Huang W 2019	NA	67/179	OR	RMA	NA	NA	≥2
Wang XC 2020	NA	33/60	OR	NA	America SC	NA	>2
Xu HR 2018	NA	30/52	OR	RMA	Santa Cruz	NA	>3
Zhang XY 2019	NA	40/56	OR	RMA	CST	NA	>3
Yoshizawa A 2010	NA	NA	OR, HR	ser2448, RMA	CST	1:100	TS3, TS4, TS5
Shimizu K 2014	80/24	33/71	OR, HR	RMA	CST, Danvers, MA	1:80	≥2
Shimizu K 2014	159/41	61/143	HR	RMA	CST, Danvers, MA	1:80	≥2
Gold KA 2014	309/61	NA	HR	NA	CST	1:100	≥2

ADC = adenocarcinoma, CP = clinicopathological, HR = hazard ratio, NA = not available, NE = negative expression, OR = odd ratio, PE = positive expression, RMA = rabbit monoclonal antibody, SCC = squamous cell carcinoma.

3.3. Quality assessment

Two researchers evaluated the quality of the 28 included studies according to the NOS scale. The quality scores varied from 6 to 8, and when the NOS score was equal to or more than 6, the articles were deemed to be of moderate to high quality. Table 4 displays the results.

Table 4.

NOS scores of included studies.

Study (year)	Selection	Comparability	Exposure/outcome	Quality scores	Grade
Dan Liu 2011	4	1	2	7	high
Zhang Y 2013	4	1	3	8	high
Oh Mee-Hye 2012	3	1	3	7	high
Lu J 2020	3	2	3	8	high
Wang L 2012	3	1	2	6	high
Kim HS 2012	3	2	2	7	high
Anagnostou 2009	4	1	2	7	high
Dhillon T 2010	4	1	3	8	high
Gately K 2012	4	1	2	7	high
Shin E 2015	3	1	3	7	high
Guo K 2014	3	2	2	7	high
Liu HB 2008	4	1	3	8	high
Xu YY 2021	3	2	2	7	high
Xu HR 2018	3	2	2	7	high
A Wang DZ 2016	3	2	2	7	high
Chen ZH 2013	3	2	2	7	high
Guo ZQ 2018	3	1	2	6	high
Hu ST 2010	3	2	2	7	high
Huang W 2019	3	1	2	6	high
Wang XC 2020	3	2	2	7	high
Xu HR 2018	3	2	2	7	high
Yuan C 2017	3	2	2	7	high
Zhang XY 2019	3	2	2	7	high
Zhang JY 2020	3	1	2	6	high
Yoshizawa A 2010	4	1	2	7	high
Shimizu K 2014	4	1	2	7	high
Gold KA 2014	4	1	3	8	high
Cai WK 2014	4	1	2	7	high

NOS = Newcastle–Ottawa scale.

3.4. Correlation between mTOR/p-mTOR expression and clinicopathological characteristics in NSCLC

In our meta-analysis, as shown in Table 5; Figures 2 and 3, we investigated the following common clinicopathological parameters: genders, ages, histological type, tumor size, depth of tumor invasion (T), lymph node metastasis (N), distant metastasis (M), differentiation, and smoking. Our analysis showed that high expression of mTOR/p-mTOR was significantly correlated with gender (OR = 0.608, 95% CI: 0.442–0.836), lymph node metastasis (OR = 2.084, 95% CI:1.437–3.022) and CEA level (OR = 1.584, 95% CI: 1.135–2.209). In contrast, the correlations were not statistically significant between mTOR/p-mTOR level and ages (OR = 0.940, 95% CI: 0.794–1.113), histological type (OR = 1.519, 95% CI: 0.896–2.576), depth of tumor invasion (OR = 1.693, 95% CI: 0.543–5.283), distant metastasis (OR = 1.425, 95% CI: 0.634–3.203), TNM stage (OR = 1.318, 95% CI: 0.906–1.918), differentiation (OR = 1.256，95% CI: 0.692–2.278, tumor size (OR = 0.908, 95% CI: 0.753–1.096), and smoking (OR = 0.719, 95% CI: 0.417–1.240). The results indicate that high expression of mTOR/p-mTOR was associated with female gender, more lymph node metastases, and higher CEA level. In addition, we analyzed the relationship between mTOR/p-mTOR and the above clinicopathological characteristics, respectively, and the results showed that the high expression of mTOR was significantly associated with more lymph node metastasis (OR = 2.074, 95% CI: 1.20–3.585); the high expression of p-mTOR was significantly associated with female gender (OR = 0.529, 95% CI: 0.313–0.896), more lymph node metastasis (OR = 2.123, 95% CI: 1.301–3.464) and more advanced TNM stage (OR = 2.049, 95% CI: 1.516–2.771) (Table 6); it coincided with the results of the above-combined analysis. All of these findings show that a high mTOR/p-mTOR expression ratio is particularly aggressive.

Table 5.

Meta-analysis of relationships between mTOR and p-mTOR expression and clinicopathological characteristics of NSCLC.

Clinicopathological characteristics	No.	No. samples			OR (95% CI)	P value	Model	Heterogeneity (I2, P)	Publication bias
		Total	PE	NE					Begg (P)	Egger (P)
Gender (Male vs Female)	20	3158	1803	1355	0.608 (0.442–0.836)	.002*	Random	I2 = 63.4%, P = .000	0.538	0.615
Age (old vs young)	18	2409	1428	981	0.940 (0.794–1.113)	.475	Fixed	I2 = 0.0%, P = .525	0.449	0.66
HT (ADC vs non-ADC)	20	3502	1982	1520	1.519 (0.896–2.576)	.121	Random	I2 = 91.1%, P = .000	0.922	0.261
N (N1-3 vs N0)	19	2793	1646	1147	2.084 (1.437–3.022)	.000*	Random	I2 = 75.8%, P = .000	0.08	0.01
TNM(ⅠⅠⅠ-Ⅳ vs Ⅰ-ⅠⅠ	11	1921	1175	746	1.318 (0.906–1.918)	.149	Random	I2 = 62.3%, P = .003	0.876	0.652
Differentiation (G3 vs G1/G2)	8	1279	761	518	1.256 (0.692–2.278)	.454	Random	I2 = 78.6%, P = .000	0.536	0.789
Tumor size (>3 vs ≤ 3)	15	2186	1310	876	0.908 (0.753–1.096)	.314	Fixed	I2 = 41.0%, P = .049	0.692	0.318
Smoking (Y vs N)	8	2050	1124	926	0.719 (0.417–1.240)	.235	Random	I2 = 84.6%, P = .000	0.536	0.305
CEA (high vs low)	4	678	337	341	1.584 (1.135–2.209)	.007*	Fixed	I2 = 45.5%, P = .138	1.000	0.690
T (3-4 vs 1-2)	3	897	588	309	1.693 (0.543–5.283)	.364	Random	I2 = 89.2%, P = .000	0.296	0.104
M (M1 vs M0)	3	786	533	253	1.425 (0.634–3.203)	.392	Fixed	I2 = 0.0%, P = .372	1.000	0.987

ADC = adenocarcinoma, CI = confidence interval, Fixed = fixed-effects model, HT = histological type, M = distant metastases, mTOR = mammalian target of rapamycin, N = lymph node metastasis, NE = negative expression, NO = reference count, OR = odds ratio, PE = positive expression, p-mTOR = phosphorylated mammalian target of rapamycin, Random = random effects model, T = depth of tumor invasion.

*p-value <0.05.

Figure 2.

Meta-analysis on the relationship between mTOR/p-mTOR expression and clinicopathological characteristics of NSCLC, including (A) genders, (B) ages, (C) histological type, (D) tumor invasion, (E) lymph node metastasis, (F) distant metastasis. CI = confidence interval, mTOR = mammalian target of rapamycin, NSCLC = non-small cell lung cancer, OR = odds ratio, p-mTOR = phosphorylated mammalian target of rapamycin.

Figure 3.

Meta-analysis on the relationship between mTOR/p-mTOR expression and clinicopathological characteristics of NSCLC, including (A) TNM stage, (B) differentiation degree, (C) tumor size, (D) smoking, (E) CEA. CI = confidence interval, mTOR = mammalian target of rapamycin, NSCLC = non-small cell lung cancer, OR = odds ratio, p-mTOR = phosphorylated mammalian target of rapamycin.

Table 6.

Relationships between mTOR/p-mTOR expression and clinicopathological characteristics of NSCLC.

Clinicopathological characteristics	mTOR					p-mTOR
	No.	OR (95% CI)	P value	Model	Heterogeneity (I2, p)	No.	OR (95% CI)	p value	Model	Heterogeneity (I2, p)
Gender (Male vs Female)	10	0.675 (0.452–1.007)	.054	Random	I2 = 50.8%, P = .032	10	0.529 (0.313–0.896)	.018*	Random	I2 = 73.1%, P = .000
Age (old vs young)	10	0.959 (0.777–1.183)	.693	Fixed	I2 = 0.0%, P = .889	8	0.908(0.684–1.206)	.506	Fixed	I2 = 39.7%, P = .114
HT (ADC vs non-ADC)	8	1.672 (0.692–4.043)	.253	Random	I2 = 91.0%, P = .000	12	1.432 (0.710–2.890)	.316	Random	I2 = 91.9%, P = .000
N (N1-3 vs N0)	10	2.074 (1.200–3.585)	.009*	Random	I2 = 79.3%, P = .000	9	2.123 (1.301–3.464)	.003*	Random	I2 = 66.6%, P = .002
TNM (ⅠⅠⅠ-Ⅳ vs Ⅰ-ⅠⅠ	6	0.989 (0.754–1.297)	.937	Fixed	I2 = 41.6%, P = .128	5	2.049 (1.516–2.771)	.000*	Fixed	I2 = 32.8%, P = .202
Differentiation (G3 vs G1/G2)	5	0.984 (0.654–1.481)	.94	Fixed	I2 = 49.2%, P = .096	3	1.613 (0.489–5.318)	.432	Random	I2 = 91.3%, P = .000
Tumor size (>3 vs ≤ 3)	8	0.798 (0.630–1.009)	.06	Fixed	I2 = 39.6%, P = .115	7	1.141 (0.835–1.559)	.407	Fixed	I2 = 32.2%, P = .182
Smoking (Y vs N)	3	0.505 (0.213–1.197)	.121	Random	I2 = 82.8%, P = .003	5	0.900 (0.443–1.830)	.771	Random	I2 = 83.9%, P = .000
CEA (high vs low)	2	1.283 (0.801–2.054)	.3	Fixed	I2 = 0.0%, P = .473	2	1.695 (0.643–4.47)	.286	Random	I2 = 71%, P = .063

ADC = adenocarcinoma, CI = confidence interval, Fixed = fixed-effects model, HT = histological type, M = distant metastases, mTOR = mammalian target of rapamycin, N = lymph node metastasis, NE = negative expression, NO = reference count, OR = odds ratio, PE = positive expression, random = random effects model, T = depth of tumor invasion.

*p-value <0.05.

3.5. Correlation between mTOR/p-mTOR expression and OS of NSCLC

3.5.1. Relationship of mTOR expression with survival.

Three articles were included in the meta-analysis of univariate analysis and four were included in the meta-analysis of multivariate analysis of the relationship between mTOR expression and prognosis. The three publications included in the meta-analysis for univariate analysis, with a total of 447 individuals, showed a pooled HR of 1.28 (95% CI: 0.65–2.51), demonstrating that there is no statistically significant association between mTOR expression and overall survival. A random effect model was adopted due to the considerable heterogeneity (P = .008, I² = 79.5%). The multivariate study comprised four publications including a total of 580 patients in the meta-analysis. Using a random effect model (P = .000, I² = 82.5%), the pooled HR was 1.62 (95% CI: 0.74–3.53). The connection between mTOR expression and prognosis remained negligible from a statistical standpoint. Table 7 and Figure 4 detail the HR and pooled HR for each research. In addition, we integrated the data from univariate and multivariate analyses to determine the pooled HR for a total of 7 data sets; the pooled HR is 1.47 (95% CI: 0.90–2.39). The results are compatible with the aforementioned conclusions.

Table 7.

Relationships between mTOR/p-mTOR expression and OS of NSCLC.

	Endpoint event	No.	HR (95% CI)	P value	Heterogeneity (I2, P)	Model
mTOR	OS	7	1.47(0.9, 2.39)	.124	81.1%, .000	Random
p-mTOR	OS	8	1.37(0.92, 2.03)	.119	81.7%, .000	Random

CI = confidence interval, HR = hazard ratio, mTOR = mammalian target of rapamycin, NSCLC = non-small cell lung cancer, OS = overall survival, p-mTOR = phosphorylated mammalian target of rapamycin.

Figure 4.

Forest plots on the prognostic significance of mTOR or p-mTOR expression for OS in NSCLC. CI = confidence interval, HR = hazard ratio, mTOR = mammalian target of rapamycin, NSCLC = non-small cell lung cancer, OS = overall survival, p-mTOR = phosphorylated mammalian target of rapamycin.

3.5.2. Relationship of p-mTOR expression with survival.

Three of the 6 papers on p-mTOR reported HR via univariate analysis, whereas 4 reported HR using multivariate analysis. Initially, a meta-analysis was conducted using data from univariate analysis, which included 756 patients with a pooled HR of 1.33 (95% CI: 0.94–1.94) The combined HR for multivariate analysis of p-mTOR was 1.42 (95% CI: 0.50–3.00), whereas the combined HR for univariate and multivariate analysis was 1.37 (95% CI: 0.92–2.03). The random effect model was adopted due to the heterogeneity. There was no statistically significant correlation between p-mTOR expression and overall survival, according to any of these studies. (Table 7; Fig. 4)

3.5.3. Relationship of mTOR and p-mTOR expression with survival.

Finally, we integrated all univariate and multivariate analyses of mTOR and p-mTOR into the meta-analysis to estimate the pooled HR to determine the association between mTOR and p-mTOR expression with OS in NSCLC patients. There were a total of 15 data sets, comprising 7 univariate analyses and 8 multivariate analyses, with a pooled HR of 1.415 (95% CI: 1.051–1.915). A random effect model was adopted due to the considerable heterogeneity between studies (I² = 80.6%, P = .001). In addition, two sets of data demonstrated a correlation between mTOR/p-mTOR expression and PFS (Progression-Free Survival) with a combined HR of 1.631 (95% CI: 0.929–2.864) using a fixed effect model (I² = 46.8%, P = .171). The statistical analysis revealed a correlation between the expression of mTOR/p-mTOR and shorter OS in patients with NSCLC, but not PFS. (Table 8; Fig. 5)

Table 8.

Meta-analysis of prognostic roles of mTOR/p-mTOR expression in NSCLC.

Endpoint event	No.	HR (95% CI)	P value	Heterogeneity (I2, P)	Model	Publication bias
						Begg (P)	Egger (p)
OS	15	1.415 (1.051–1.905)	.022*	80.6%, .000	Random	0.843	0.54
PFS	2	1.631 (0.929–2.864)	.089	46.8%, .171	Fixed	1.000	/

CI = confidence interval, HR = hazard ratio, mTOR = mammalian target of rapamycin, NSCLC = non-small cell lung cancer, OS = overall survival, PFS = progression-free survival, p-mTOR = phosphorylated mammalian target of rapamycin.

*p-value <0.05.

Figure 5.

Meta-analysis on the prognostic significance of mTOR and p-mTOR expression for OS, PFS in patients with NSCLC. CI = confidence interval, HR = hazard ratio, mTOR = mammalian target of rapamycin, NSCLC = non-small cell lung cancer, OS = overall survival, PFS = progression-free survival, p-mTOR = phosphorylated mammalian target of rapamycin.

In terms of the link between mTOR/p-mTOR and OS, the independent and combined studies produced contradictory results. No statistically significant link was found between mTOR or p-mTOR and OS, regardless of whether univariate or multivariate analysis was included. We investigated the causes of this discrepancy. First, although the included studies satisfied the inclusion and exclusion criteria, the quality of individual research varied, resulting in heterogeneity within studies and disparities between the results of grouped and combined analyses. Secondly, the meta-analysis of mTOR or p-mTOR alone comprised a small number of studies, a small sample size of the study population, and high heterogeneity, which rendered the pooled conclusions unstable. Upon synthesis of all investigations, the sample size increased, resulting in statistically significant findings. In addition, the grouping analysis was not based on randomized comparisons, and the probability of false negative/false positive significance tests rose without modifying the significance test level; hence, the results’ dependability is questionable. Lastly, p-mTOR is the phosphorylated version of mTOR and the downstream protein of mTOR; both serve the same role in the control of autophagy, therefore we feel that combining the two studies is compatible with its mechanism, hence increasing the sample size and the reliability of the results.

3.6. Subgroup analyses on the prognostic value of mTOR/p-mTOR expression for OS

To further assess the predictive importance of mTOR/p-mTOR, we conducted a stratified analysis; the findings are presented in Table 9. All of the included studies were divided into subgroups based on area, sample size, statistical analysis, cutoff values, gender, and histological type. Stratifying by geographic region, the pooled HR for studies done in China was 1.787 (95% CI: 1.013–3.15), indicating that mTOR/p-mTOR expression was substantially related to poor prognosis; however, the outcome was different in the non-China subgroup (HR = 1.212, 95% CI: 0.886–1.646). The association between mTOR/p-mTOR expression and poor prognosis remained statistically significant in subgroups with fewer ADC patients (HR = 1.825, 95% CI: 1.322–2.521) and subgroups with fewer males (HR = 1.563, 95% CI: 1.098–2.224). mTOR/p-mTOR expression was not statistically related to prognosis in any category when categorized by statistical analysis method and cutoff value. In conclusion, the findings of the subgroup analysis revealed that the positive expression of mTOR/p-mTOR increased the probability of mortality in Chinese female squamous carcinoma patients, giving a rationale for the use of mTOR inhibitors in certain groups.

Table 9.

Subgroup analyses for the relationship between mTOR/p-mTOR expression and OS in NSCLC.

Subgroups	No.	HR (95% CI)	P value	Heterogeneity (I2, P)
By statistical analysis
Univariate analysis	7	1.314 (0.976–1.767)	.071	66.3%, .007
Multivariate analysis	8	1.517 (0.859–2.682)	0.151	87.0%, .000
By region
China	6	1.787 (1.013–3.15)	0.045*	83.6%, .000
non-China	9	1.212 (0.893–1.646)	0.218	71.6%, .000
By No. of patients
<200	12	1.575 (1.128–2.200)	0.008*	76.4%, .000
≥200	3	0.963 (0.643–1.443)	0.857	69.5%, .038
By male (%)
<70	9	1.563 (1.098–2.224)	.013*	79.9%, .000
≥70	6	1.214 (0.709–2.080)	.479	79.9%, .000
By cutoff values
≥2	7	1.523 (0.849–2.735)	.159	88.9%, .000
non-≥2	8	1.330 (0.998–1.774)	.052	60.1%, .014
By ADC (%)
<50	10	1.825 (1.322–2.521)	.000*	71.0%, .000
≥50	5	0.893 (0.642–1.241)	.499	60.3%, .039

ADC = adenocarcinoma, CI = confidence interval, HR = hazard ratio, mTOR = mammalian target of rapamycin, NSCLC = non-small cell lung cancer, OS = overall survival, p-mTOR = phosphorylated mammalian target of rapamycin.

*p-value <0.05.

3.7. Sensitivity analysis

A sensitivity analysis was conducted to determine the dependability of the aggregated results. We assessed the reliability and consistency of the pooled data by analyzing the impact of excluding each study on the overall results. Among the clinicopathological parameters, seven studies with substantial variability were analyzed for sensitivity, including gender, histological type, tumor invasion (T), lymph node metastasis (N), TNM stage, differentiation, and smoking status. In the examination of each clinicopathological feature, excluding any one research did not affect the pooled OR. As shown in Figure 6, the pooled results of OS did not change after excluding studies with substantial heterogeneity, such as Yoshizawa, A. et al and K Shimizu (HR = 1.489, 95% CI: 1.052–2.108), showing that the results were consistent.

Figure 6.

Sensitivity analysis on the prognostic value of mTOR/p-mTOR expression for OS in patients with NSCLC. CI = confidence interval, mTOR = mammalian target of rapamycin, NSCLC = non-small cell lung cancer, OS = overall survival, p-mTOR = phosphorylated mammalian target of rapamycin.

3.8. Publication bias

Begg’s funnel plot and Egger’s s bias indicator test were used to investigating publication bias. In none of the included research was a substantial publication bias identified (Tables 5 and 7).

4. Discussion

From yeast to humans, autophagy is a highly conserved mechanism dependent on the activity of a core group of autophagy-related proteins.^[80] mTOR was identified as the homologous gene of rapamycin protein target (TOR) in yeast mutants in 1991.^[8] mTOR binds to multiple proteins to form two distinct signaling complexes, mTORC1 (comprised of mTOR, Raptor, mLST8, PRAS40, and DEPTOR) and mTORC2 (comprised of mTOR, Rictor, mLST8, mSin1, Hsp70, and DEPTOR), which trigger different downstream signals to regulate cellular function by binding to specific substrates.^[81] mTORC2 facilitates phosphorylation of AKT at Ser473, therefore activating mTORC1; active mTORC1 promotes ribosomal protein S6 kinase (S6K), which drives the mTORC2 complex. In addition, mTORC1 is sensitive to energy and pressure and strongly inhibited by rapamycin, whereas mTORC2 is insensitive to rapamycin and nutrition,^[82,83] but its activity is inhibited by long-term long-lasting rapamycin treatment.^[84] mTOR signaling plays a central role as a central controller in the energy supply for cell survival, growth, metabolism, proliferation, and death, such as the classical PI3K/AKT/mTOR signaling pathway being highly activated in several tumor types, regulating several cellular mechanisms, transmitting signals from upstream regulatory proteins such as PTEN, PI3K, and RTKs, as well as many downstream effectors, such as GSK-3, FOXO, and MDM2.^[85] In addition, the Liver kinase B1/AMP-activated protein kinase/mTOR pathway works as a molecular characterization of the tumor suppressor axis, suggesting a connection between cancer and metabolism.^[86] The aforementioned molecular mechanisms reveal that mTOR may activate and phosphorylate diverse stimuli and signals via several signaling pathways to govern the pathophysiology of cancer. The expression of mTOR/p-mTOR is increasingly recognized as being associated with a variety of malignancies. Aberrant activation of mTOR signal caused by genetic or metabolic disorders can significantly promote tumor growth, angiogenesis, and metastasis.^[87] Additional signal components of the upstream and downstream mTOR pathways are also altered during proliferation disorders, which is associated with the poor prognosis of cancers.^[88]

In recent years, several clinical studies have investigated the relationship between the positive expression of mTOR/p-mTOR and numerous prevalent malignancies, such as NSCLC. This meta-analysis collected a wide number of clinicopathological parameters, examined both mTOR and p-mTOR expression, and employed univariate and multivariate analyses. Secondly, no publication bias was identified in the study, and sensitivity analysis revealed no statistically significant differences when omitting any articles. Positive expression of mTOR/p-mTOR was related to females, more lymph node metastases, and higher CEA levels, which was partly consistent with the findings of Xin-Chen Wang et al^[89] We hypothesize that the clinicopathological relevance of mTOR/p-mTOR expression may partially explain its prognostic role in NSCLC. However, more clinical studies with larger samples are required to determine if mTOR is an observable indicator for the development of NSCLC. Moreover, while utilizing mTOR inhibitors, the combination of personalized medications can be considered after stratification according to the clinicopathological features of patients, therefore minimizing drug resistance, enhancing effectiveness, and enhancing prognosis. Further research is required to determine the precise molecular mechanism behind the relationship between mTOR and other clinicopathological parameters.

Notably, the meta-analysis revealed that the expression of mTOR/p-mTOR was related to a shorter OS in NSCLC patients. The strong connection remained statistically significant in subgroups stratified by geographic region, sample size, gender, and histological type. All of these pooled analyses indicate that mTOR/p-mTOR may be an indicator of a poor prognosis in NSCLC. If the HR is larger than 2, prognostic indicators are deemed to have a significant predictive value for a poor prognosis; however, the HR in this study is less than 2. As evidence remains limited, the validity of these findings may be compromised by very short follow-up periods and small sample sizes. In addition, the accuracy of the overall pooled results of OS based on univariate analysis may be marginally diminished by unremoved confounding variables that may influence the prognosis of NSCLC, including TNM stage, tumor invasion, differentiation, and LNM. The determination of the real involvement of mTOR and p-mTOR in NSCLC may be hindered by these potential factors. Therefore, based only on the findings of 13 retrospective observational studies, the current findings are insufficient for drawing definitive conclusions. The influence of mTOR/p-mTOR activation on the prognosis of NSCLC should be explored further in long-term studies with bigger sample numbers and more extensive follow-ups.

In addition, the pooled HR for overall survival revealed a correlation between the expression of mTOR and p-mTOR with the prognosis of NSCLC. However, when the association between mTOR/p-mTOR and survival was evaluated independently, the results indicated that the expression of mTOR/p-mTOR was not statistically significant with NSCLC patients’ prognosis. This contradiction may be a result of the small number of original studies included in this meta-analysis. In addition, when stratified by geographical region, gender, and histological type, the link between mTOR/p-mTOR expression and poor prognosis was maintained in the Chinese subgroup, which included a greater number of female patients and fewer individuals with ADC. The prognostic significance of mTOR/p-mTOR expression in various malignancies has been extensively researched in the past. Xian-Fei Ding et al^[90] revealed that the overexpression of p-mTOR in breast cancer was not substantially connected with the prognosis of OS and disease recurrence. In head and neck cancers, it has been suggested that mTOR pathway proteins can be employed as survival predictors for patients with head and neck cancer, given that their expression was substantially connected with poor OS and short DFS.^[91] In research on gastric cancer, Hua Wang et al found that mTOR overexpression significantly predicts a bad prognosis. In contrast, the cumulative survival rate of patients with mTOR expression was considerably greater in Xiao et al^[92] research than that of patients lacking mTOR expression. In addition, a study on nasopharyngeal cancer revealed that the overexpression of mTOR predicted poor overall survival and disease-free survival in patients with nasopharyngeal carcinoma.^[93] Therefore, no conclusion has been made about the predictive importance of mTOR/p-mTOR in cancer.

There are several limitations in our meta-analysis. First, despite a comprehensive literature search, the number of studies on the relationship between mTOR/p-mTOR expression and prognosis was too small, and this portion of the data was also excluded because extraction of survival data from the K-M curves may produce inaccurate results,^[94] so the pooled HRs may be overestimated or underestimated due to the small and incomplete raw data. Only one of the included articles examined the association between mTOR/p-mTOR expression and PFS, only two data sets were studied, and the findings of the meta-analysis were inconclusive. In addition, unknown confounding factors in the original studies may have influenced the outcomes of the pooled HRs. Secondly, in the pooled analysis of clinicopathological features, the grouping criteria of some studies were contradictory with those of other research, making their inclusion difficult. The heterogeneity of the studies included in some statistical indicators was high, but the stratified analysis was impossible to undertake due to insufficient information in the original articles. Thirdly, although all studies used IHC to detect mTOR/p-mTOR expression, the proportion of positive samples varied widely, ranging from 38.3% to 77.1%, suggesting that the level of homogeneity of included studies may be affected by the assessment criteria of staining, cutoff values, antibodies and dilutions of experimental reagents, and the level of laboratory personnel, which may be an important source of deviation. Even while no significant publishing bias was discovered, it is still possible that bias exists, given there is a general inclination to report significant positive outcomes.

5. Conclusions

In conclusion, our meta-analysis revealed a correlation between the positive expression of mTOR/p-mTOR and gender, lymph node metastasis, and CEA level. mTOR/p-mTOR may also be predictive of a poor outcome in NSCLC. However, numerous difficulties and limitations have not been sufficiently addressed in this meta-analysis. More high-quality research based on multivariate analysis with large sample size is required to validate and modify our existing findings.

Author contributions

Conceptualization: Weiwei Qiu, Meiying Ren.

Data curation: Yuhua Fu, Yan Liu.

Formal analysis: Weiwei Qiu, Meiying Ren.

Investigation: Cuifeng Wang, Yuhua Fu, Yan Liu.

Methodology: Weiwei Qiu, Meiying Ren, Cuifeng Wang.

Software: Weiwei Qiu, Meiying Ren.

Writing – original draft: Weiwei Qiu, Yan Liu.

Writing – review & editing: Meiying Ren, Weiwei Qiu.