Two librarians developed the search strategy using the PRESS guidelines to cross-check results.8 Medline via PubMed (29/01/2023), Embase (29/01/2023), and Cochrane CENTRAL (29/01/2023) were searched from inception using free and MeSH term synonyms of “malignant pleural effusion” and “indwelling pleural catheter” plus filters for SRs, randomized controlled trials (RCTs), and cohorts (see supplementary material for the detailed search strategies).

Studies were eligible if they included adult patients with MPE secondary to any metastatic tumor or primary pleural tumors such as mesotheliomas, irrespective of the SCT (chemotherapy, immunotherapy, molecular targeted therapies, or combinations) instituted, and whether or not the lung was trapped. In addition, to be eligible, the intervention for MPE should be ambulatory, and the outcome should have been recorded considering the time of IPC placement in relation to SCT (before, during, or after SCT). Studies with radiotherapy as the only cancer treatment were excluded, as were those in which the results of treated and untreated arms could not be derived separately. Concerning the type of studies, RCTs, quasi-controlled trials, prospective and retrospective cohorts, and case series involving over 20 patients were all eligible. SRs covering the same topic were screened for additional primary studies that had not yet been included. The lists of references of the included studies were likewise scanned for potential further studies that would have been overlooked.

The studies retrieved by the searches were added to an Endnote 21™ (Clarivate Analytics) library and uploaded to the Rayyan® systematic review software (https://www.rayyan.ai/), where a pair of reviewers performed the study selection independently. First, duplicates were removed, after which the titles and abstracts were screened to discard unrelated studies. All remaining reports were read in full, and data were collected.

For each study, the eligibility criteria were checked upfront; if these were not met, the study was added to the list of excluded studies, with a comment specifying the unmet criteria.

Data collected from the primary studies were transcribed by one of the reviewers into an Excel® template, which were agreed a priori and cross-checked by the other reviewer. The variables to be collected were related to: (1) study: design, year, country, setting (single center or multicenter), follow-up duration; (2) samplecharacteristics: sample size, cancer type, demographic characteristics (age and gender), and performance status (Karnofsky, Eastern Cooperative Oncology Group); (3) lung and pleural effusion characteristics: expandable/non-expandable lung, pleural loculations, and fluid data (malignant or non-malignant cells, pH, cellular predominance); (4) cancer treatment: use of chemotherapy, immunotherapy, molecular targeted therapy or combinations, and lines of therapy; (5) IPC: indication (symptomatic relief, pleurodesis), drainage schedule (on demand, daily), and catheter implantation setting (inpatient or outpatient); (6) outcomes of interest: efficacy (symptom improvement, need for further procedures, total number of procedures or re-interventions required, hospital stay, patient survival, degree of auto-pleurodesis achieved [different definitions], and time to catheter removal); safety (intra-pleural infection or empyema, and extra-pleural infection including cellulitis, exit site and tunnel tract infections; symptomatic loculations, metastasis of the tract, pain, bleeding, or other complications such as obstruction).

For each outcome, the definition, effect measure, crude, and adjusted values, as well as covariates were collected, and results were expressed by timing groups. After plotting all the results in tables, the researchers met in order to discuss visualization of results and aggregability of studies.

To assess the risk of bias within the included studies, the Cochrane Risk of Bias (RoB) tool for RCTs and Risk of Bias In Non-randomized Studies of Interventions (RoBINS-I) for non-randomized studies were employed. Studies were assessed by a single reviewer and then cross-checked by another.

Meta-analysis was only planned if the studies were sufficiently homogeneous. A random-effects (DerSimonian and Laird) method for meta-analysis was used. Specifically, a meta-analysis on the risk of IPC-related infections from studies which separated data by groups (with and without SCT while IPC was in situ) was performed. Heterogeneity was estimated from an inverse-variance fixed-effect model and presented as I2. The Stata 12.0 (Stata Statistical Software: Release 12. College Station, TX: StataCorp LLC) metaprop and metan commands were used.

Survival was expressed as median survival time, 6-month mortality, and mortality with IPC in situ. All studies demonstrated a more prolonged survival in those patients in whom the IPC was inserted close to SCT or during SCT.11,13,15–19

Akram et al. reported that patients who received chemotherapy before and after IPC insertion exhibited longer median survival in comparison with those who did not receive any oncological treatment (106 vs 41 days; p=0.004).11 Consequently, the decision regarding institution of the oncological treatment is an essential factor contributing to survival. According to Hak et al. the 6-month mortality was lower in patients with IPC and concomitant SCT than in those without SCT (35% vs 59%; p=0.007).13 The results of Morel et al. pointed toward the same direction, with lower mortality rates in patients undergoing SCT upon IPC insertion compared with those who did not receive SCT (57% vs 68%), although there was no mention of whether this difference was statistically significant.16 The study by Ost et al. demonstrated an improvement in quality-adjusted survival in patients starting chemo- or radiotherapy immediately after IPC insertion.17 Mitchell et al. showed that chemotherapy in IPC carriers increased survival time (238 vs 114 days),15 defined as the median time from IPC insertion to death. Finally, Whilshire et al. observed that SCT within a month of IPC insertion or during IPC use was associated with a 16% increase in survival (hazard ratio [HR]=0.84; p=0.015).19

Conducting a meta-analysis was not considered, given the variability of the outcomes and definition of the interventions.

Death and auto-pleurodesis were the most common causes of IPC removal.12,15–18,20 The rate of spontaneous pleurodesis was high, ranging between 23% and 50% (Table 5). Although using concomitant chemotherapy when the IPC was in situ did not impact probability of removal in two studies,15,18 Wang et al. observed greater rates of spontaneous pleurodesis in patients with lung cancer who were subjected to epidermal growth factor receptor (EGFR) therapy while the IPC was in situ (adjusted OR=3.87)20 (Table 5).

On the other hand, information on the duration of the IPC in situ was not fully consistent.12–16,18–20 Although the variability of the studies does not allow drawing solid conclusions, the results suggested longer catheter duration in patients undergoing SCT, except for the Hak et al. study.13 However, the differences were only statistically significant in the Mekhaiel et al. study14 and could be related to longer survival with chemotherapy. Concerning the remaining studies, multivariate analyses either did not reveal differences between patients with or without SCT upon IPC placement15 or the statistical significance of the comparison was not recorded.13,16,18,19

Performing a meta-analysis was not taken into consideration, owing to the difficulty of pooling median times and interquartile ranges, which were the most reported effect measures.

Three studies presented data about additional pleural procedures being required. Overall, the frequency of pleural re-interventions was low, but the results were mainly descriptive, without any comparison between patients with or without chemotherapy while the IPC was insitu.12,17,18 Nevertheless, in the study by Gilbert et al.,12 where all 91 patients included underwent chemotherapy, no additional thoracocentesis was reported, and there was only one placement of a chest tube (1.1%), and 3 thoracoscopic pleurodesis procedures (3.3%) (Table 5).

All the included studies presented data on IPC-related infections, whether intrapleural (empyema) or extrapleural (cellulitis, exit site infection or line tract infection). Nevertheless, the direct effect of SCT and the precise time of infection during treatment were not properly evaluated by all the authors.

The overall infection prevalence was highly variable, which can probably be accounted for by methodological differences among studies. Overall, IPC-related infections ranged from 1.9%15 to as high as 25.4% in one publication,11 but in the latter the authors did not specify the infection sites. By infection location, the presence of empyema as a complication of IPC ranged from 1%17 to 8%,12 while infections pertaining to the extra-pleural space (cellulitis) exhibited less variation, from 0.7%14 to 4%.17 The pooled estimated rate of IPC-related infections was 2.85% (95% confidence interval: 2.24–3.45; I2 73.39%) (Fig. 2S).

Two of the 10 included studies did not present data on the possible relationship between infection occurrence and SCT timing.11,20 The Akram et al. study reported a significant reduction (p=0.03) in the infection rate of patients receiving domiciliary care education, but the infection frequency was not analyzed in relation with the chemotherapy timing.11 Wang et al. reported a prevalence of intra-pleural infection of 1.9%, also without information on its possible relationship with SCT timing.20 The association between complications and SCT timing was analyzed globally, clearly demonstrating that SCT with concurrent or sequential radiotherapy following IPC insertion did not at all increase the overall risk of complications. The association between infection and SCT timing in relation to IPC insertion was analyzed in eight of the 10 studies.12–19 None of them found a significant increase in infection risk associated with the SCT timing, despite the presence of neutropenia12–19 (Table 6). A meta-analysis showed that the pooled relative risk (RR) of IPC-related infections while on SCT compared to off-therapy was 0.98 (95% CI: 0.93–1.03; I2 0.0%) (Fig. 3S).

Other outcomes comprised IPC-related pain and drain blockage, without any effect of SCT timing,13 in addition to symptomatic loculations, metastasis of the tract, leaks, or tumor seeding which were not analyzed with respect to SCT.18,20

The review protocol was registered in PROSPERO, prior to the search strategy, under number CRD42022322026.

The Endnote® libraries and data extraction forms are available upon reasonable request.