Intrathoracic infusion therapy of thymic peptides and chemical irritants for malignant pleural effusion: a meta-analysis of 24 randomized controlled trials


 Background: To further determine the clinical efficacy and survival of intrathoracic infusion with TPs and chemical irritants and their therapeutic threshold and optimal control regimen to achieve a desired response in malignant pleural effusion (MPE). Methods: We collected all randomized controlled trials (RCTs) of TPs with chemical irritants from Chinese and English databases (from inception until September 2019), and performed a new meta-analysis following the PRISMA guidelines. We measured their bias risk, summarized data using meta-analysis, and evidence quality using the Grades of Recommendation Assessment, Development and Evaluation approach. Results: We collected 24 trials involving three TPs and platinum and 1,592 patients. Most trials had unclear bias risk. TPs with platinum significantly improved complete response [4.02 (3.12 to 5.18)] and quality of life [3.64 (2.34 to 5.66)], the 0.5-year overall survival (OS) rate [5.75 (3.02 to 10.92)], and 1-year OS rate [5.29, (1.71 to 16.36)], and reduced the treatment failure, myelosuppression, and gastrointestinal toxicity. For patients with moderate to large volume of pleural effusion, KPS score ≥50 to 60, or AST ≥3 months, the thymosin (200–300mg/time), thymopentin (2mg/time) or thymosin alpha 1 (3.2mg/time) with cisplatin (30–40mg/m 2 ), carboplatin (400mg/m 2 ), or oxaliplatin (100mg/m 2 ) are possible regimens for achieving a desired success, and low failure. Most results were robust and moderate quality. Conclusion: The moderate evidence suggests that the TPs with platinum is beneficial to the patient with MPE, and provides evidence for the therapeutic threshold and possible regimens that may achieve a desired success and reduce the failure.


Background
Malignant pleural effusion (MPE), often originates from most malignant tumors, is a common and challenging problem. Patients often suffer from breathlessness, poor quality of life (QOL) and prognosis [1][2]. Control of the effusion signi cantly reduces morbidity, and improves quality of life. Chemical pleurodesis is a rst-line intervention for MPE without a nonexpendable lung (NEL) [3]. After draining effusion, a chemical irritant such as talc poudrage [4], tetracycline [5], bleomycin [6], cisplatin (DDP) [7], doxycycline [8], or silver nitrate [9], among others is instilled into the pleural cavity to induce intrapleural in ammation and brosis, and then improve clinical symptoms. However, intrathoracic infusions of chemical irritants remain palliative, with a median survival ranging from 3 to 12 months [10][11]. Thus, developed new control strategies are utterly important.
Biological response modi ers (BRMs) have purported important biological activities as anti-tumor, anti-infection, and immunomodulation [12][13]. As important BRMs, thymic peptides (TPs) included the puri ed thymus extracts (pTE) as thymosin extracted from thymus, and the synthetic thymic peptides (sTP) as thymosin alpha 1 (Tα1) and thymopentin. Tα1, a 28-amino acid peptide, and thymopentin, a ve-peptide were rst described and characterized from calf thymuses in 1977 [14] and 1979 [15][16], respectively. Owing to their pleiotropic biological activities, TPs has been used in the treatment of cancers and infectious diseases in clinic [17][18][19][20]. Our previous systematic review and meta-analysis [21] found that the sTP, especially Tα1 with chemotherapy might improve anti-tumor immunity, tumor responses, QOL, and 1-year overall survival (OS) rate, and result in a low incidence rate of neutropenia, thrombocytopenia and gastrointestinal reactions. However, can intrathoracic infusion therapy of TPs and chemical irritants improve clinical e cacy and survivals of the patients with MPE? A previous meta-analysis [22] reported that thymosin combined with oxaliplatin might improve the clinical response and host immunity, and decrease the incidence of adverse drug reactions (ADRs) in lung cancer with MPE. However, they did not conclusively determine whether administration of the TPs with chemical irritants improves the clinical response, survival, and reduces the ADRs. Additionally, their therapeutic threshold and optimal control regimen to achieve a best possible clinical response and desired safety have yet to be determined. So far, new original studies [23][24][25] have been published. Therefore, the aim of this meta-analysis was to further determine the clinical e cacy and survival of intrathoracic infusion with TPs and chemical irritants, and their therapeutic threshold and optimal control regimen to achieve a best possible clinical response and desired safety, and to provide an optimal evidence for developing an individualized control strategy based on TPs and chemical irritants for MPE.

Materials And Methods
Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [26], we designed, performed and reported a metaanalysis about intrathoracic infusion therapy of TPs and chemical irritants in MPE.

Inclusion And Exclusion Criteria
All eligible studies must meet following criteria. All subjects had malignant tumors with MPE, which was diagnosed following thorax imaging, pleural uid analysis, cytology, or pleural biopsy. All patients had no limitations on the primary tumors, and thoracentesis or indwelling pleural catheters (IPCs). Before intrathoracic infusion therapy, all patients had normal liver, kidney, and heart function. The intervention studied was TPs administered via intrapleural infusion, and not the subcutaneous, intravenous, or intramuscular injection. The experimental group was administered TPs with chemical irritants, and the control group was administered chemical irritants alone. One month before therapy, no patients received intrapleural infusion using chemical irritants, other BRMs, traditional Chinese medicine (TCM), or hyperthermia. The primary outcomes were treatment success, treatment failure, and survival, and the secondary outcomes were QOL, peripheral blood lymphocytes, ADRs, and thoracentesis-related adverse events (TRAEs). All studies were randomized controlled trials (RCTs), and no restrictions on follow-up protocols or research institutions.
All un-eligible studies must meet the following criteria: duplicates; studies of non MPEs, and non TPs therapy; studies of TPs alone or plus other BRMs, TCM, hyperthermia, or systemic chemotherapy; meeting abstracts, reviews without speci c data, and irrelevant systematic reviews or meta-analyses; cohort, case control, single arm studies, or case reports; studies without data on treatment success, treatment failure, survival, peripheral blood lymphocytes, QOL, ADRs, or TRAEs.

Study Selection
According to the pre-designed inclusion and exclusion criteria, two reviewers (You-Shu Shen and Yuan-Xiu He) independently collected all eligible studies about intrathoracic infusion therapy of TPs and chemical irritants in MPE. Any disagreements were resolved by discussing themselves or with a third investigator (Zheng Xiao).

Evaluation Of Methodological Bias Risk
According to the Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [27], the parameters for methodological bias risk were as follows: random sequence generation, allocation concealment, blinding, incomplete outcome data, selective reporting, and other bias (e.g., whether the baseline was comparable). Two reviewers (Xiao-Tian Zheng and Ya-Hui Yang) independently evaluated each trial, judged each item as three levels (Yes for a low bias, No for a high risk of bias, and unclear). Any disagreements about judgment of bias risk were solved by discussion between themselves or with a third investigator (Zheng Xiao).

Outcome Measures
The primary outcomes were treatment success, treatment failure, and survival. We integrated previous criteria [28 -33] as follows: (1) complete response (CR) is a pleural effusion disappeared for more than four weeks, or the lack of accumulation of uid; (2) partial response (PR) is a pleural effusion was reduced more than 50% for more than four weeks or a 50% decrease in the effusion; (3) no response (NR) or stable disease (SD) is pleural effusion was reduced less than 50% or increased less than 25% or the effusion recurred but required no further therapy; (4) progressive disease (PD) is pleural effusion increased more than 25% along with other signs of progressive disease or symptomatic re-accumulation of the effusion requiring repeat thoracentesis or chest tube. We de ned the treatment success as CR plus PR, and treatment failure as NR/SD plus PD. The survival was evaluated using overall survival (OS) rate, progression-free survival (PFS) rate, or hazard ratio (HR) of the OS and PFS.
Secondary outcomes were QOL, peripheral blood lymphocytes, ADRs, or TRAEs. First, the QOL was considered improved, when the scores increased ten points or higher after treatment according to the Karnofsky Performance Status (KPS) Scale [34][35]. Second, the peripheral blood lymphocytes were represented as the proportions of CD3 + , CD3 + CD4 + , and CD3 + CD8 + T cells and ratio of CD4 + /CD8 + T cells, which were detected using ow cytometry or indirect immuno uorescence test after treatment. Third, according to the World Health Organization (WHO) [36] or Common Terminology Criteria for Adverse Events (CTCAE) standards [37], the ADRs were represented as myelosuppression, gastrointestinal toxicity (gastrointestinal reaction and nausea/vomiting), hepatotoxicity (serum aminotransferase or alkaline phosphatase > 1.25 × N), and nephrotoxicity (serum urea nitrogen or creatinine > 1.25 × N), fever, and chest pain. Fourth, the TRAEs were represented as respiratory failure, catheter-related infection/chest infection, pneumothorax, and cutaneous emphysema, among others.

Data Extraction
According to the pre-designed data extraction form, the extracted data were the time of publication; the patient baselines as the primary tumors, volumes of pleural effusion, KPS score, anticipated survival time (AST), IPCs, and treatment process (primary treatment (PT)/retreatment(RT)); cases of experimental and control group; demographic and methodological characteristics; types and usages of TPs and chemical irritants; follow-up protocols; evaluation criteria; and outcomes including treatment success, treatment failure, OS, PFS, QOL, ADRs, TRAEs, and peripheral blood lymphocytes. Two reviewers (Min He and Shu-Guang Li) independently extracted all data. If the studies without Kaplan-Meier survival curves or other relevant data, we contacted their authors to gain available survival data. If the authors unavailable, we re-constructed the Kaplan-Meier survival curves into OS, PFS, and HR using Engauge Statistical Analysis And Quality Of Evidence We represented the treatment success, treatment failure, OS rate, PFS rate, QOL, ADRs, TRAEs as odds ratio (OR) and 95% con dence intervals (CI); OS and PFS as HR and 95% CI; and peripheral blood lymphocytes as standardized mean difference (SMD) and 95% CI. We applied the Cochran's χ 2 test and the I 2 statistic to measure potential statistical heterogeneity. I 2 > 50% meant a statistical heterogeneity. If p > 0.1 and I 2 ≤ 50%, we summarized the OR, HR, SMD, and their 95% CI using a xed-effects model. If I 2 > 50% and without obvious clinical heterogeneity, we summarized the data using a random-effects model.
Otherwise, we eliminated the indicator. Two reviewers (Cheng-Qiong Wang and Teng-Yang Fan) performed a series of meta-analyses using Review Manager 5.3 (The Cochrane Collaboration, Oxford, UK). Included studies > 10, we examined potential publication bias using funnel plot and Egger/Begg's tests. Following our experiences [21,[40][41][42], if the trial with at least one domain considered as high risk, we de ned it as poor quality. In this meta-analysis, and if the result was statistically different and bene t to TPs infusion, we de ned the trial as under-or over-estimated trial. Then, we developed a sensitivity analysis model to further examine the results' robustness under extreme conditions as removing the trials with poor quality, under-estimated ADRs, or over-estimated e cacy.
Following previous guidance [43], we developed a subgroup analysis model to examine the heterogeneity and reveal the effects of variables as patient baselines (primary tumors, volume of pleural effusion, KPS score, AST, treatment process, and IPCs), interventions, controls and evaluation criteria on treatment success or treatment failure between different studies. We developed a univariable random effects meta-regression for the relationship between each variable and treatment success or treatment failure, and also a post-hoc multiple regression analysis adjusting for the OR of treatment success and treatment failure under all variables.
Following the Grades of Recommendation Assessment, Development and Evaluation (GRADE) approach [44] and our experiences [21,[40][41][42], we sorted evidence quality into four grades as high, moderate, low and very low. The criteria of quality degradation were as follows: (1) limitations in study design (trials had unclear risk, and some trials had high risk, if sensitivity analysis was good robust, the evidence was downgraded by one level; if poor, the evidence was downgraded by two levels; all trials had no high risk, and the evidence was downgraded by one level; and if all were poor, the evidence was downgraded by two levels.); (2) statistical heterogeneity (the statistical heterogeneity were shown among the trials, and sensitivity analysis was poor robust); (3) indirectness, (the subjects, interventions, controls, or outcomes did not meet this study); (4) imprecision (the subjects for each outcome < 300); (5) publication bias (the indicator showed a publication bias, and sensitivity analysis was poor robust.). For (2) to (5), the evidence was downgraded by one level. According to the above criteria, two reviewers (Cheng-Qiong Wang and Xiao-Fan Chen) applied GRADE pro ler to summarize the evidence quality, and generate absolute estimates of effect for the treatment success, treatment failure, OS, PFS, QOL, ADRs, TRAEs, and peripheral blood lymphocytes.

Characteristics Of Included Trials
This meta-analysis collected 24 eligible trials [23][24][25]45-65] with 1,592 patients and three TPs and platinum for analysis (Table 1). Patient ages ranged from 27 to 78 years old, and 911 and 627 cases were male and female, respectively. The experimental group was administered TPs with platinum in 798 cases, and the control group was administered platinum alone in 794 cases. Seventeen trials described 1214 patients with lung cancer [23-25,   were used with one to two times per week for one to eight times. The indicators were measured after treatment four to 12 months. Twenty-four trials with 1,592 patients [23][24][25]45-65] all reported the treatment success and failure, and two trials with 226 patients [57,62]

Clinical Response
Twenty-four trials [23][24][25] with 1,592 patients reported the treatment success/ failure ( Fig. 3a and 3b). Cochran's χ 2 test and I 2 statistic did not nd statistical heterogeneity in treatment success and failure (I² = 0%). Therefore, the data of treatment success and failure was summarized by using a xedeffects model. Compared with platinum alone, the result of meta-analysis demonstrated that intrathoracic infusion of

Overall Survival
Two trials with 226 patients [57,62] reported the OS rates (Fig. 4). There was no heterogeneity in 0.5-or 1-year OS rates (I² = 0%). Therefore, the data of OS rate was summarized by using a xed-effects model. The results demonstrated that TPs with platinum signi cantly increased the 0.
Cochran's χ 2 test and I 2 statistic found minimal heterogeneity in myelosuppression (I 2 = 14%), and no heterogeneity in other ADRs or TRAE (I 2 = 0%). Therefore, the data of ADRs and TRAEs was summarized by using a xed-effects model. The results demonstrated that TPs with platinum resulted in a lower incidence rate of myelosuppression [OR = 0.46, 95% CI (0.31 to 0.69), p = 0.0001), gastrointestinal toxicity [OR = 0.46, 95% CI (0.33 to 0.64), p < 0.00001) than platinum alone. However, no statistical differences were shown in chest pain, fever, hepatotoxicity, and nephrotoxicity. Additionally, two trials with 136 patients [47,51] reported that no TRAEs had occurred in both groups.

Subgroups And Meta-regression Analysis
Patient baselines were the primary tumors, volume of pleural effusion, KPS score, and AST. First, primary tumors included malignant tumors and lung cancer.
The results of subgroup analysis demonstrated that intrathoracic infusion of TPs with platinum signi cantly improved the treatment success, and resulted in a low risk of failure for MPE from lung cancer and malignant tumors (Table.3a, Fig.S8 and S10). According to volume of pleural effusion, MPE was mainly moderate to large. TPs with chemical platinum signi cantly improved the treatment success, resulted in a low risk of failure (Table.3b, Fig.S12 and S14). KPS scores were ≥ 50, and ≥ 60, and AST was mainly ≥ 3 months. Intrathoracic infusion of TPs also achieved the above effects in patients with KPS scores ≥ 50 or ≥ 60 (Table.3c, Fig.S16 and S18), or AST ≥ 3 months (Table.3d, Fig.S20 and S20). Univariable meta-regression did not nd any correlation between treatment success / failure and any of baselines (p = 0.93 for primary tumor, p = 0.67 for volume of pleural effusion, p = 0.71 for KPS, and p = 0.70 for AST) ( Table.3 and Fig.S9, S11, S13, S15, S17, S19, S21, and S23).  (Table.3e and 3f, Fig.S24, S26, S 28, and S30). The TPs with platinum was used with once or twice a week, and it achieved the above effects (Table.3 g, Fig.S32 and S34). Treatment lasting one to eight times, mainly three to four times, it still did so (Table.3 h, Fig.S36 and 38). The evaluation criteria were Millar and Ostrowskimj [28][29][30][31][32][33]. The TPs with platinum achieved the above effects using the two criteria (Table.3i, Fig.S40 and 42). Univariable meta-regression showed no correlation between treatment success/failure and variables (p = 0.99 for TPs, p = 0.85 for platinum, p = 0.96 for treatment frequency, and p = 0.73 for treatment times, and p = 0.93 for criteria) (Table.3 and Fig.S25, S27, S29, S31, S33, S35, S37, S39, S41 and S43). Finally, multiple meta-regression analysis did not nd any correlation between treatment success/failure and all variables (Table 3).

Sensitivity Analysis
In this meta-analysis, 10 trials with poor quality [45-51, 53, 56, 63] were shown in treatment success, treatment failure, QOL, myelosuppression, gastrointestinal toxicity, fever, TRAEs, CD3 + T cells, CD3 + CD4 + T cells, and CD4 + /CD8 + T cells ratio. The results of sensitivity analysis demonstrated that all results were robust before and after rejecting the poor trials (Table 4a-4b). The trials might over-estimate the treatment success, QOL, 0.5-year OS rate, 1-year OS rate, CD3 + T cells, CD3 + CD4 + T cells, and CD4 + /CD8 + T cells ratio, and under-estimate the treatment failure, myelosuppression, and gastrointestinal toxicity. The results of sensitivity analysis demonstrated that except for the 0.5-year and 1-year OS rate, other results had robustness before and after rejecting the trials with over-estimated e cacy or under-estimated ADRs (Table 4c-4d). In all, most results were robust.
Statistical heterogeneity was shown in CD3 + T cells, CD3 + CD4 + T cells, and CD4 + /CD8 + T cells ratio, and sensitivity analysis showed good robustness; and no publication bias was shown in treatment success, treatment failure, myelosuppression, and gastrointestinal reaction; therefore, we did not downgrade their quality. The sample size for 0.5-year-, one-year OS rate, hepatotoxicity, nephrotoxicity, and TRAEs were less than 300 patients, therefore, we downgraded their quality with one level. No outcomes were eligible for upgrade. Taken together, the quality of evidence was very low for TRAEs; low for 0.5-year, one-year OS rate, hepatotoxicity, nephrotoxicity; and moderate for others (Table.5).

Discussion
With the exception of adjuvant therapy for malignant tumors, intrathoracic infusion of TPs with chemical irritants is also used in the control of MPE. In this meta-analysis, we included 24 eligible trials [23-25, 45-65] with 1,592 patients to determine whether they improve the clinical e cacy and survivals.
Compared with platinum alone, the results of meta-analysis demonstrated that TPs in combination with platinum signi cantly improved the treatment success, QOL, 0.5-year and 1-year OS rate, and reduced the incidence of treatment failure, myelosuppression, and gastrointestinal toxicity in MPE. Most trials only reported ADR, and ignored potential TRAEs. In addition, they signi cantly increased the proportions of CD3 + T cells, CD3 + CD4 + T cells, and the ratio of CD4 + /CD8 + T cells. In methodology, most had unclear bias risk, and 10 poor [45-51, 53, 56, 63] failed in completely reporting the ADRs. The peripheral blood lymphocytes had statistical heterogeneity, the CD3 + CD8 + T cells had obvious clinical heterogeneity, and we summarized the data using a random-effects model. As a result of limited trials, we gave up exploring the reason of heterogeneity using subgroup analysis. The trials objectively reported the treatment success, failure, myelosuppression, and gastrointestinal toxicity. The 0.5-year and 1-year OS rate were poor robust. With the exception of the anemia, thrombocytopenia, TRAEs, 0.5-year, one-year OS rate, hepatorenal toxicity, all outcomes had moderate quality based on the GRADE approach.
A meta-analysis [22] reported that thymosin combined with oxaliplatin might improve the clinical response, and host immunity, and with low incidences of the ADRs in lung cancer patient with MPE. However, the meta-analysis [22] had many shortcomings in the methodology, and an increasing new clinical trials have reported on the evaluation of TPs with platinum. This meta-analysis improved the design, integrated previous meta-analysis [22], and supplemented 18 trials with 1,132 patients [23-25, 45-55, 58, 61, 63, 65], and further demonstrated that intrathoracic infusion of TPs with platinum signi cantly improved the treatment success, 0.5-year, one-year OS rate, and with low incidences of the treatment failure, myelosuppression and gastrointestinal toxicity. In addition, we found that they signi cantly up-regulated the level of peripheral blood lymphocytes. Our related systematic review and meta-analysis [21] found that the sTP, especially Tα1with chemotherapy might improve the anti-tumor immunity, tumor response, QOL and 1-year OS rate of lung cancer. Another related study [66] reported that the thoracic injection of low-dose interleukin-2 signi cantly improved the clinical response and QOL of patients with MPE. Oka M,et.al [67] reported that an important BRM, Lentinan signi cantly improved clinical response and augmented lymphokine activated killer activity through intrathoracic infusion for patients with malignant peritoneal and/or pleural effusions. These results provided the indirect evidence of our ndings. Based on this metaanalysis ndings and evidence quality, we believe that intrathoracic infusion of TPs with platinum may improve the treatment success, QOL, 0.5-year, one-year OS rate, up-regulate the level of peripheral blood lymphocytes, and with low incidences of the treatment failure, myelosuppression and gastrointestinal toxicity.
But, these results may under-estimate the ADRs and TRAEs (Fig. 7).
In 24 trials, patient baselines, types and usages of TPs and platinum had marked diversity. Hence, we carried out a series of subgroup analyses to reveal the optimal condition of TPs with platinum. The subgroup analysis found that the TPs with platinum all signi cantly improved the treatment success, and with a low risk of failure in patients with moderate to large volume of pleural effusion, KPS scores ≥ 50 or ≥ 60, or AST ≥ 3 months. Chemical pleurodesis is considered by many as a rst-line intervention in MPE without a NEL, particularly in cases with an AST of > 3 months [68-69]. Additionally, Yoon DW et al [70] reported that the KPS score (≥ 50 to 60) was independent predictor of a poor survival after video-assisted thoracoscopic surgery (VATS)-mediated talc pleurodesis. Compared with VATS-mediated talc pleurodesis, the TPs with platinum seems to have a lower threshold. These ndings indicate that moderate to large volume of pleural effusion, KPS score ≥ 50 to 60, or AST ≥ 3 months may be the therapeutic threshold for TPs with platinum infusion. Millar or Ostrowskimj criteria and the type of primary tumor had no negative impact on these results. Whether it also does for primary treatment/retreatment or drug-resistant patients remain unclear. The subgroup analysis further found that intrathoracic infusion of thymosin (200-300 mg/time), thymopentin (2 mg/time) or Tα1 (3.2 mg/time) with DDP (30-40 mg/m 2 ), CBP (400 mg/m 2 ), or L-OHP (100 mg/m 2 ), all signi cantly improved the treatment success, and with a low risk of failure. The TPs with platinum was used once or twice a week, mainly for 3-4 times. Interestingly, the thymosin, thymopentin or thyamlfasin in combination with DDP, CBP, or oxaliplatin all signi cantly improved the treatment success, with a low risk of failure. However, univariable and post-hoc multiple regression analysis did not nd any positive correlation, and these conclusions from the subgroup analysis belonged to indirect evidence. As a result of limited trials, this meta-analysis didn't demonstrate the optimal TPs type, usage and combination with platinum for achieving the best response, and the relationship between TPs and platinum also needs to be con rmed by new trials.
Based on the optimization of success and failure, we believe that the moderate to large volume of pleural effusion, KPS score ≥ 50 to 60, or AST ≥ 3 months may be the therapeutic threshold for TPs with platinum infusion. The thymosin (200-300 mg/time), thymopentin (2 mg/time) or Tα1 (3.2 mg/time) with DDP (30-40 mg/m 2 ), CBP (400 mg/m 2 ), or L-OHP (100 mg/m 2 ) are possible regimens for producing a desired treatment success, and low failure. The optimal type, usage and combination with platinum for achieving the best response need to be con rmed by new trials. If successful, these ndings will be important for de ning therapeutic threshold, developing rational usage of TPs and platinum, and attaining an individualized TPs intrathoracic infusion therapy for MPE (Fig. 7).
There were some shortcomings in this meta-analysis. First, we only retrieved Chinese and English databases, which might have led to retrieval bias. Second, most trials did not describe the volume of pleural effusion, KPS score, AST, and treatment process. Third, only ten trials [23-24, 48-49, 53, 55, 57, 61, 64-65] described random sequence generation, and ten trials [45-51, 53, 56, 63] failed in completely reporting the ADRs, and most ignored potential TRAE and PFS.
Fourth, their quality was moderate to very low. Univariable and multiple meta-regression analysis did not nd any positive correlation. Fifth, there was lack of a uni ed standard for clinical response. As a result of limited trials, this meta-analysis didn't demonstrate the optimal type, usage and combination with platinum for achieving the best response. All of these shortcomings might have led to insu cient evaluation of the outcomes.

Conclusions
The evidence indicates that intrathoracic infusion of TPs with platinum may improve the treatment success, QOL, 0.5-year, and one-year OS rate, up-regulate the level of peripheral blood lymphocytes, and with low incidences of the treatment failure, myelosuppression and gastrointestinal toxicity. The moderate to large volume of pleural effusion, KPS score ≥ 50 to 60, or AST ≥ 3 months may be the therapeutic threshold for TPs with platinum infusion. The thymosin       Note: Thoracentesis-related adverse events (TRAEs), TNF-α: tumor necrosis factor-alpha, SM: statistical method, FEM: xed-effects model, REM: randomeffects model, OR: odds ratio, SMD: standardized mean difference, CI: con dence interval; Poor trials (Poor*) had at least one domain considered as high risk of bias; Over* or Under*: over or underestimated trials which results were signi cant difference and bene cial to TPs perfusion.  The analysis of OS rates between the two groups Meta-analysis results of peripheral blood lymphocytes Figure 6 The analysis of publication bias