Placebo effect in the treatment of non-small cell lung cancer: a meta-analysis : Journal of Bio-X Research

Secondary Logo

Journal Logo

Research Articles

Placebo effect in the treatment of non-small cell lung cancer: a meta-analysis

Ren, Siyuana,b; Ma, Mengyaoa,b; He, Chuana,b; Deng, Yuhuia,b; Chen, Xiaoyunc; Liu, Yonglind; Jin, Yangyange; Liu, Yansongf,*; Cai, Leia,b,*; He, Lina,b

Author Information
doi: 10.1097/JBR.0000000000000123

Abstract

Introduction

Lung cancer is a debilitating disease with the highest morbidity and mortality rates worldwide. Non-small cell lung cancer (NSCLC) is the most common type of lung cancer, accounting for ~85% of new cases.[1] The majority of patients are diagnosed when the disease has progressed to advanced or metastatic stages (IIIB or IV) and prognosis is estimated to be less than 1 year.[2] For cases of advanced NSCLC, chemotherapy and immunotherapy have significant advantages over surgery and radiotherapy in terms of symptom control and prolonging survival. However, the questionable efficacy and inevitable adverse effects have resulted in a plateau in clinical practice.[3,4] Effective novel therapeutic strategies are required to prolong survival times and improve the quality of life of patients.

Palliative care, as defined by the World Health Organization, is a specific type of treatment aiming to prevent and relieve patients’ suffering by means of impeccable identification and estimation of physical, psychosocial, and spiritual problems associated with certain life-threatening diseases.[5] Palliative care may activate placebo effects in the body through focusing on the management of symptoms and psychosocial support.[6] These effects are actually the responses of the human body to the general expectancy caused by receiving certain physical or psychological cues.[7]

Many large-scale clinical phase III drug tests for NSCLC are randomized controlled trials (RCTs) incorporating a placebo group as control to identify the effects of new drugs.[8-10] The outcomes of placebo-treated patient groups in RCTs should provide key information on placebo effects and provide further indications of the effects of palliative care. To evaluate the potential significance of palliative care for NSCLC, we performed both single-arm and two-arm meta-analyses on the therapeutic and adverse responses to placebo in clinical NSCLC drug RCTs using either Bayesian random-effects or conventional models.

Materials and methods

This meta-analysis was conducted in agreement with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.[11]

Literature review and selection

Literature published from January 1, 1978 to September 1, 2020 was searched using several main electronic databases, specifically, PubMed (MEDLINE), Scopus, Web of Science, and China National Knowledge Infrastructure. The following keywords were used: (lung OR pulmonary) AND (cancer OR carcinoma OR neoplasm OR tumor) AND placebo AND non-small cell. The title, abstract, and full text of each paper were methodically retrieved and subsequently reviewed independently by MM and YD according to previously described procedures.[12-15] The reference lists were also manually checked for additional studies.

Inclusion and exclusion criteria

The final eligible papers included for statistical analysis met the following criteria: (1) RCT was the study design, (2) the cancer type was solely NSCLC stage III or IV, (3) phase II or phase III pharmaceutical human studies were conducted, (4) complete data regarding drug/placebo effects, adverse effects, and primary endpoint were available, and (5) Jadad scores were ≥4.[16] Studies were excluded if they were: (1) non-RCTs, (2) abstracts, reports, or papers from conferences, (3) reviews, letters and commentaries, (4) duplicate or multi-duplicate samples, or (5) lacking data on drug/placebo effects/adverse effects. For duplicate studies, the most recent publication was retained; while for duplicate samples, the study with the largest sample size was retained.

Data extraction

The following general information was extracted from each study: (1) the number of either active drug-treated or placebo-treated subjects, (2) treatments for drug and placebo groups, (3) the specific type and progression stage of NSCLC, and (4) the route of drug or placebo administration. Additional information extracted from each study and finally incorporated in the meta-analysis included the number of subjects in the drug or placebo group under the categories of therapeutic effects, adverse effects, and consequences of adverse effects, that is, dropout and dose reduction.

Following the Response Evaluation Criteria In Solid Tumors (RECIST) criteria,[17] the therapeutic effects of either active drug or placebo were classified into the following groups: (1) complete response (CR), whereby every treated lesion disappeared and the short axes of treated nodes were all reduced to no more than 10 mm, (2) partial response (PR), whereby total diameters of all target lesions achieved a reduction of at least 30% compared with the baseline, (3) progressive disease (PD), whereby the sum of diameters of target lesions reached both a >20% increase compared with the minimum in the study and an absolute increase of 5 mm or higher, (4) stable disease (SD) that did not qualify for CR, PR, or PD, and (5) controlled disease (CD) representing the sum of CR, PR, and SD. Adherence to treatment was classified into discontinuation caused by adverse effects and dose reduction caused by adverse effects. Adverse effects included rash, anorexia or inappetence, nausea, vomiting, aspartate aminotransferase elevation, alanine aminotransferase elevation, fatigue, hypertension, thyroid-stimulating hormone elevation, diarrhea, pruritus, and dyspnea, among others. The primary endpoints for each study were progression-free survival (PFS) and overall survival (OS).

Quality assessment

A quality score for each study was calculated following the Jadad method to assess whether the information provided was sufficient.[16] The score (five points in total) was based on the assessment of three aspects of a clinical study: randomization (two points), double-blinding (two points), and disposal of dropouts and withdrawals (one point). The score assigned for each item was dependent on its completeness and appropriateness. Furthermore, the Cochrane risk of bias tool was used to assess literature quality from seven perspectives: (1) random sequence generation, (2) allocation concealment, (3) blinding of participants and personnel, (4) blinding of outcome assessment, (5) incomplete outcome data, (6) selective reporting and (7) other bias.[18] Risk of bias assessments were judged independently by MM, CH, and YD. Disagreements in the review process were resolved based on decisions by a third-party comprising CH and LC.

Statistical analysis

A Bayesian random-effects model combined with a binomial-normal hierarchical model was used in this study, since this method accommodates more heterogeneity than the classic random model.[19] The 95% credible intervals (CrI) for the Bayesian method and 95% confidence intervals (CI) for the conventional method were calculated to indicate that the real value had 95% probability of being within the interval. The analysis was performed using Fast*Pro Software (Medical College Hospitals Philadelphia, USA).[19,20] Review Manager (RevMan Version 5.3, Cochrane Community, London, UK) was additionally utilized to analyze the data using a random-effects model and heterogeneity between studies evaluated with I2. I2>50% was considered statistically significant and pooled effects were calculated according to the random-effects model. Otherwise, when I2<50%, pooled effects were evaluated according to the fixed-effects model. G*Power software (University of Dusseldorf, Dusseldorf, Germany)[21] was used for calculating the statistical power to detect effects.

Results

Literature screening and characteristics of included studies

Eligible studies were screened according to the PRISMA guidelines (Fig. 1).[22] In total, 2198 papers were retrieved through electronic database searching and reference list checking. After the removal of duplicates, 719 studies remained. Based on the inclusion and exclusion criteria, 18 studies were identified for qualitative review (Additional Table 1, https://links.lww.com/JR9/A39). Among these, only five studies included either 0.9% sodium chloride or best supportive care in the placebo group with no mention of active drugs while the others had low-dose active drugs in the placebo group. Accordingly, these five studies incorporating 2245 drug-treated and 1510 placebo-treated patients were selected for the meta-analysis.[8-10,23,24] The results are summarized in Table 1. The quality assessment results of these five studies according to the Jadad method are listed in Table 2. Low risk of bias was observed for all five included studies (Fig. 2).

Table 1 - Overview of the included studies
Adverse effects [n (%)] Dropout [n (%)] Dose reduction caused by adverse events [n (%)] Therapeutic response [n (%)] Primary end point
Studies Type Drug group Placebo group Drug group Placebo group Drug group Placebo group Type Drug group Placebo group Drug group Placebo group
Shepherd et al, 2005[24] Fatigue 383 (79.0) 179 (74.0) 24 (5) 5(2) 92 (19) 5(2) CR 3 (0.7) Median PFS: 2.2 months Median PFS: 1.8 months
Anorexia 335 (69.1) 136 (56.2) PR 40 (8.2) 2 (0.8)
Nausea 194 (40.0) 82 (33.9) Median OS: 6.7 months Median OS: 4.7 months
Vomiting 121 (24.9) 56 (23.1)
Rash 369 (76.1) 41 (16.9)
Diarrhea 55 (11.3) 19 (3.9)
Ciuleanu et al, 2009[23] Fatigue 108 (24.5) 23 (10.4) 21 (5) 3 (1) 22 (5) 2 (1) CR+PR+SD 228 (52) 74 (33) Median PFS: 4.3 months Median PFS: 2.6 months
Nausea 83 (18.8) 12 (5.4)
Anorexia 82 (18.6) 11 (5.0) Median OS: 13.4 months Median OS: 10.6 months
Vomiting 39 (8.8) 3 (1.4)
Rash 9 (2.0) 2 (0.9)
Diarrhea 23 (5.2) 6 (2.7)
Lee et al, 2012[8] Diarrhea 287 (46.4) 34 (11.2) 75 (12.1) 16 (5.3) 161 (26) 14 (4.6) PR 16 (2.6) 2 (0.7) Median PFS: 1.9 months Median PFS: 1.8 months
Rash 262 (42.3) 33 (10.9)
Fatigue 110 (17.8) 51 (16.8) Median OS: 8.5 months Median OS: 7.8 months
Vomiting 84 (13.6) 38 (12.5)
Nausea 139 (22.5) 52 (17.2)
Cappuzzo et al, 2010[10] All 281 (64.9) 89 (20.0) 20 (5%) 7 (2%) 70 (16%) 15 (3%) CR+PR 52 (11.9) 24 (5.4) Median PFS: 12.3 weeks Median PFS: 11.1 weeks
Rash 258 (59.6) 34 (7.6) CR+PR+SD 178 (40.8) 122 (27.4)
Pruritus 27 (6.2) 9 (2.0) Median OS: 12.0 months Median OS: 11.0 months
Diarrhea 79 (18.2) 14 (3.1)
Anorexia 22 (5.1) 10 (2.2)
Cicènas et al, 2016[9] All 255 (79.2) 181 (56.7) 10 (3.1%) 3 (0.9%) 45 (14%) 12 (3.8%) CR 3 (0.9) 2 (0.6) Median PFS: 13.0 weeks Median PFS: 12.0 weeks
No details PR 18 (5.6) 10 (3.1)
SD 176 (54.7) 178 (55.5)
CR+PR+SD 197 (61.2) 190 (59.2)
CD=controlled disease, CR=complete response, DFS=disease-free survival, OS=overall survival, PD=progression disease, PFS=progression-free survival, PR=partial response, RFS=relapse-free survival, SD=stable disease.

Table 2 - Quality assessment based on Jadad method
Items Shepherd et al, 2005[24] Ciuleanu et al, 2009[23] Lee et al, 2012[8] Cappuzzo et al, 2010[10] Cicenas et al, 2016[9]
Was the study described as random? 1 1 1 1 1
Was the randomization scheme described as appropriate? 1 1 1 1 1
Was the study described as double-blind? 1 1 1 1 1
Was the method of double blinding appropriate? 1 1 1 1 1
Was there a description of dropouts and withdrawals? 1 1 1 1 1
Score 5 5 5 5 5

F1
Figure 1.:
Flowchart of the study selection. RCTs = randomized controlled trials.
F2
Figure 2.:
Risk of bias evaluation using the Cochrane method. (A) Risk of bias summary; (B) Risk of bias of each study.

Therapeutic effects

All five studies provided data regarding CR, PR, and SD in a separate or combined manner based on the RECIST criteria. To determine the therapeutic response to placebo, the total CD rate in the placebo group was calculated as 24.1% (95% CrI, –0.126 to 0.609) according to the Bayesian model of single arm meta-analysis (Fig. 3).

F3
Figure 3.:
Forest plot for Bayesian meta-analysis of the response rate in control groups of NSCLC drug RCTs. The quoted estimate is shown in black and the shrinkage estimate in gray. Crl=credible interval, NSCLC=non-small cell lung cancer, RCTs=randomized controlled trials.

Survival analyses were further performed using PFS and OS data from the five studies including evaluation of the active drugs erlotinib, vandetanib, and pemetrexed. The heterogeneity among the five RCTs was not significant (P=0.55 and P=0.66, respectively; I2=0 for both) and risk ratio (RR) was calculated using a fixed-effects model. Significant differences in PFS and OS were observed between placebo and active drug groups (P=0.02 and P=0.007, respectively; Fig. 4). Compared with the active drug group, RR values of 0.81 (95% CI, 0.68–0.97) and 0.85 (95% CI, 0.76–0.96) for PFS and OS, respectively, were calculated for the placebo group (Fig. 4).

F4
Figure 4.:
Forest plot for meta-analysis of survival of patients in five NSCLC drug RCTs. (A) Progression-free survival; (B) Overall survival. CI=confidence interval, NSCLC=non-small cell lung cancer, RCTs=randomized controlled trials.

Adherence to treatment

Since temporary disruption of treatment was seldom reported, only the rates of dropout and dose reduction caused by adverse effects were analyzed. The dropout rate of placebo treatment was 2.1% (95% CrI, 0.007–0.039) and the dose reduction rate was 3.0% (95% CrI, 0.017–0.045). We observed significant differences between placebo and active drug groups in terms of dropout and dose reduction rates, with RR=0.39 (P<0.00001) and RR=0.24 (P<0.0001), respectively (Fig. 5). The statistical power analysis for the Mantel-Haenszel test indicated 81.7% power to derive the RR estimate of 0.3.

F5
Figure 5.:
Forest plot for meta-analysis of outcomes caused by adverse effects. (A) Dropout rate; (B) Dose reduction rate. CI=confidence interval.

Adverse events

Four out of the five included studies provided detailed information on adverse effect types. Common adverse effects reported in NSCLC drug trials were fatigue, anorexia, nausea, vomiting, diarrhea, and rash. The results of a separate meta-analysis on each of these common placebo-related adverse events are summarized in Table 3. Among the adverse events, fatigue was the most frequent, with an incidence of 37.3% (–0.545 to 1.291).

Table 3 - Meta-analysis of the most common adverse events in the placebo groups of NSCLC clinical trials
Types Percentage (95% CI)
Fatigue 37.3% (−0.545–1.291)
Anorexia 23.6% (−0.607–1.079)
Nausea 20.1% (−0.38–0.787)
Vomiting 13.1% (−0.368–0.635)
Rash 9.4% (−0.08–0.274)
Diarrhea 6.2% (−0.038–0.157)
CI=confidence interval, NSCLC=non-small cell lung cancer.

Discussion

NSCLC contributes to a leading proportion of all cancer-related deaths and the 5-year overall survival rate remains less than 10% for stage IV patients.[25] In recent years, palliative care has been proposed as an alternative treatment modality for patients with advanced or metastatic stage NSCLC to prolong survival times and is believed to exert therapeutic effects through activation of placebo effects within the human body.[6] The placebo effect is widely considered to be a result of psychophysiological changes caused by the symbolic significance of treatment rather than specific pharmacological or physiological features.[26] The opposite of placebo is the nocebo effect, whereby patients are convinced that treatment is noneffective, and therefore, therapy without specific curative effects makes the patients’ condition worse. To evaluate the significance of palliative care, which faces diverse ethical challenges, the effects of placebo treatment in NSCLC new drug RCTs were analyzed.

Double-blinded RCT is the gold standard for clinical tests of new drugs, in which 0.9% sodium chloride or a sugar pill is used as placebo. However, due to ethical issues, a small dosage of active drug is often included in placebo treatment. In our analysis, 18 studies were screened for qualitative review, among which only five either used 0.9% sodium chloride only or solely mentioned placebo as the placebo treatment group. To specifically evaluate the placebo effects indicative of palliative care, these five studies were included for quantitative analysis.

In the present analysis, the CD rate of placebo treatment was determined as 24.1% and 95% CI ranged from –0.126 to 0.609. Our findings suggest that the effects of placebo or palliative care cannot be precisely estimated to alleviate symptoms and control disease progression. However, in 2003, a totally different response rate (ranging from 2% to 7%) following placebo treatment was reported in oncology trials.[27] The response rates of NSCLC to placebo treatment require further investigation.

We further estimated the effects of placebo on PFS and OS compared with the active drug treatment groups (erlotinib, vandetanib, and pemetrexed) in NSCLC RCTs. The results ofour meta-analysis showed that PFS and OS risk were significantly decreased (0.81 and 0.85; P=0.02 and 0.007, respectively) in the placebo group. However, 95% CI was so narrow that for some cases, the placebo effects on PFS and OS appeared similar to those of active drugs (95% CI 0.68–0.97 and 0.76–0.96, respectively). Our findings suggest that placebo exerts similar effects to the active drugs tested in terms of prolonging survival time, supporting the theory that personalized palliative care with the best support could potentially achieve reasonable therapeutic effects.

The rate of dose reduction and dropout due to adverse effects with placebo treatment was estimated to be 2% to 3%. These results are similar to the previous estimation that 4% to 26% of patients discontinued placebo treatments for different diseases due to perceived side effects and suggest moderate or mild toxicity caused by adverse events in placebo treatment groups.[28] Compared with active drug administration, placebo treatment had a significant RR of 0.39 and 0.24 for dropout and dose reduction, respectively. Based on the collective findings, we suggest that palliative care offers an advantage in improving patient quality of life compared with treatments with active drugs, such as erlotinib, vandetanib, and pemetrexed, although the incidence of adverse effects in the placebo group remains to be established.

There are several limitations in the present analysis. One is the relatively small number of trials included. Only five studies meet the inclusion criteria, and further studies incorporating more eligible trials are required. The other is that we cannot exclude a selection bias in original trials, so that trial participants have an overall better survival than nonparticipants. So, in RCTs, more strict measures are suggested to avoid the selection bias.

In summary, data of dropout, dose reduction rates, and adverse events from the current study indicate that placebo has low toxicity in NSCLC patients. Notably, toxicity caused by placebo is significantly lower than that by the active drugs, such as erlotinib, vandetanib, or pemetrexed, in NSCLC drug RCTs. Although the therapeutic responses to placebo treatment could not be determined here, survival analyses suggest that palliative care exerts comparable effects to active drugs in terms of prolonging survival time. Further research is warranted to fully elucidate the benefits of personalized palliative care in prolonging survival in NSCLC patients.

Acknowledgments

The authors are grateful to Dr. David M. Eddy, Vic Hasselblad, and David Behar (Medical College Hospitals Philadelphia, USA) for providing Fast*Pro Software.

Author contributions

LC, YoL, and LH participated in study design, literature search, data analysis, manuscript writing, and review. SR participated in manuscript writing and review. MM participated in literature search, sample, and data collection. CH, YD, and YJ participated in literature search and data collection. XC and YaL participated in review. All authors approved the final version of the manuscript.

Financial support

This study was supported by grants from the Natural Science Foundation of Shanghai (No. 19ZR1427700), the Ministry of Science and Technology of China (No. 2017YFC1001302), Shanghai Key Laboratory of Psychotic Disorders (13dz2260500) in Shanghai Mental Health Center, China (No. 19-K02). The content is solely the responsibility of the authors and does not represent the official views of the funding agencies.

Conflicts of interest

The authors have no conflicts of interest to disclose.

Editor note: LH is an Editorial Board member of Journal of Bio-X Research. He was blinded from reviewing or making decisions on the manuscript. The article was subject to the journal’s standard procedures, with peer review handled independently of this Editorial Board member and their research groups.

References

[1]. World cancer report: World Health Organization 2020. https://publications.iarc.fr/586
[2]. Hopwood P, Stephens RJ. Depression in patients with lung cancer: prevalence and risk factors derived from quality-of-life data. J Clin Oncol. 2000;18:893–903.
[3]. Brahmer J, Reckamp KL, Baas P, et al. Nivolumab versus docetaxel in advanced squamous-cell non-small-cell lung cancer. N Engl J Med. 2015;373:123–135.
[4]. Reck M, Rodriguez-Abreu D, Robinson AG, et al. Pembrolizumab versus chemotherapy for PD-L1-positive non-small-cell lung cancer. N Engl J Med. 2016;375:1823–1833.
[5]. Borasio GD. Translating the World Health Organization definition of palliative care into scientific practice. Palliat Support Care. 2011;9:1–2.
[6]. Ferris FD, Bruera E, Cherny N, et al. Palliative cancer care a decade later: accomplishments, the need, next steps - from the American Society of Clinical Oncology. J Clin Oncol. 2009;27:3052–3058.
[7]. Cai L, He L. Placebo effects and the molecular biological components involved. Gen Psychiatr. 2019;32:e100089.
[8]. Lee JS, Hirsh V, Park K, et al. Vandetanib versus placebo in patients with advanced non-small-cell lung cancer after prior therapy with an epidermal growth factor receptor tyrosine kinase inhibitor: a randomized, double-blind phase III trial (ZEPHYR). J Clin Oncol. 2012;30: 1114–1121.
[9]. Cicènas S, Geater SL, Petrov P, et al. Maintenance erlotinib versus erlotinib at disease progression in patients with advanced non-small-cell lung cancer who have not progressed following platinum-based chemotherapy (IUNO study). Lung Cancer. 2016;102:30–37.
[10]. Cappuzzo F, Ciuleanu T, Stelmakh L, et al. Erlotinib as maintenance treatment in advanced non-small-cell lung cancer: a multicentre, randomised, placebo-controlled phase 3 study. Lancet Oncol. 2010;11:521–529.
[11]. Moher D, Shamseer L, Clarke M, et al. Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015 statement. Syst Rev. 2015;4:1.
[12]. Cai L, Huang T, Su J, et al. Implications of newly identified brain eQTL genes and their interactors in schizophrenia. Mol Ther Nucleic Acids. 2018;12:433–442.
[13]. Lei C, Dongqing Z, Yeqing S, et al. Association of the CTLA-4 gene with rheumatoid arthritis in Chinese Han population. Eur J Hum Genet. 2005;13:823–828.
[14]. Cai L, Wan CL, He L, et al. Gestational influenza increases the risk of psychosis in adults. Med Chem. 2015;11:676–682.
[15]. Cai L, Cai MH, Wang MY, et al. Meta-analysis-based preliminary exploration of the connection between ATDILI and schizophrenia by GSTM1/T1 gene polymorphisms. PLoS One. 2015;10: e0128643.
[16]. Jadad AR, Moore RA, Carroll D, et al. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Control Clin Trials. 1996;17:1–12.
[17]. Therasse P, Arbuck SG, Eisenhauer EA, et al. New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst. 2000;92:205–216.
[18]. Cai L, Deng SL, Liang L, et al. Identification of genetic associations of SP110/MYBBP1A/RELA with pulmonary tuberculosis in the Chinese Han population. Hum Genet. 2013;132:265–273.
[19]. Garmendia Madariaga A, Santos Palacios S, Guillén-Grima F, et al. The incidence and prevalence of thyroid dysfunction in Europe: a metaanalysis. J Clin Endocrinol Metab. 2014;99:923–931.
[20]. Llavero-Valero M, Guillén-Grima F, Zafon C, et al. The placebo effect in thyroid cancer: a meta-analysis. Eur J Endocrinol. 2016;174: 465–472.
[21]. Faul F, Erdfelder E, Lang AG, et al. G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods. 2007;39:175–191.
[22]. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med. 2009;6:e1000097.
[23]. Ciuleanu T, Brodowicz T, Zielinski C, et al. Maintenance pemetrexed plus best supportive care versus placebo plus best supportive care for non-small-cell lung cancer: a randomised, double-blind, phase 3 study. Lancet. 2009;374:1432–1440.
[24]. Shepherd FA, Rodrigues Pereira J, Ciuleanu T, et al. Erlotinib in previously treated non-small-cell lung cancer. N Engl J Med. 2005;353:123–132.
[25]. Goldstraw P, Chansky K, Crowley J, et al. The IASLC lung cancer staging project: proposals for revision of the TNM stage groupings in the forthcoming (Eighth) edition of the TNM classification for lung cancer. J Thorac Oncol. 2016;11:39–51.
[26]. Brody H. Placebos and the Philosophy of Medicine: Clinical, Conceptual and Ethical Issues. Chicago: University of Chicago Press; 1980.
[27]. Chvetzoff G, Tannock IF. Placebo effects in oncology. J Natl Cancer Inst. 2003;95:19–29.
[28]. Kaptchuk TJ, Miller FG. Placebo effects in medicine. N Engl J Med. 2015;373:8–9.
Keywords:

lung cancer; meta-analysis; NSCLC; palliative care; placebo effect

Supplemental Digital Content

Copyright © 2022 The Chinese Medical Association, Published by Wolters Kluwer Health, Inc.