Literatures searching results

PubMed, Embase, Web of science, and Cochrane Library were used to search for publicly available RCTs, and a total of 3977 relevant literature were retrieved; excluding duplicates, 3476 articles remained. By reading the titles and abstracts, 53 articles were left. By reading the full text, a total of 8 papers were finally included for meta-analysis (Fig. 1).

Fig. 1

Flowchart for inclusion of literature

Basic information about the included literature

The 8 included studies [8, 13,14,15,16,17,18,19] were published in English between 2000 and 2020, with a total sample size of 4240 cases, including 2548 cases in the experimental group and 1692 cases in the control group. The main study population of 3 studies published by Pere, Jürgen and James R, were NSCLC patients, and 4 studies published by Sylke, Robert, Hye-Suk and Thomas were SCLC patients; Johan’s main study population included NSCLC patients and SCLC patients. Four studies used darbepoetin alfa, 3 studies used epoetin alfa, and 1 study used RCHT+EPO. By organizing and summarizing the baseline data of the 8 included studies, the baseline characteristics were basically balanced between the experimental and control groups. Following is the basic information and basic characteristics of the included studies (Tables 1 and 2).

Table 1 Basic information on included studiesTable 2 Basic characteristics of included studiesResults of risk of bias assessment

An assessment of the methodological quality of the included studies is shown in Table 3. For the quality assessment of the literatures, most of the assessed entries were identified as low risk, which suggested no significant risk of bias (Fig. 2).

Table 3 Methodological quality assessment of included studiesFig. 2

Risk of bias evaluation of the included literature

Data analysesMortality

A total of 6 included studies (I2 = 40%, P = 0.14) reported mortality of patients from the start of treatment to the end of study follow-up, with 1892 (71.10%) of 2661 patients in the erythropoietin group and 1320 (72.17%) of 1829 patients in the control group. The result showed that the risk of mortality was lower in patients using ESAs than the control group (RR 0.96, 95% CI 0.92–0.99, P = 0.02) (Fig. 3).

Fig. 3

Forest plot of total mortality between erythropoietin and control groups

We removed data of the Pere 2020 which had a larger sample size, and there was no significant difference in mortality risk between patients treated with ESAs and control groups (RR 0.99, 95% CI 0.92–1.06, P = 0.69) (Fig. 4).

Fig. 4

Forest plot of total mortality between the erythropoietin group and the control group after exclusion of the literature with a larger proportion of the sample size

By using Stata 14.0 statistical software for sensitivity analysis of the 5 included studies, the main indicators of each group were excluded one by one analysis, and the final results did not change much, indicating that the present results are reliable (Fig. 5).

Fig. 5

Mortality sensitivity analysis

Among the included studies, 2 studies included patients with NSCLC and 2 included SCLC, so we did subgroup analyses based on lung cancer type. And the results showed no significant difference in the risk of mortality between the NSCLC subgroup (RR 1.01, 95% CI 0.87–1.17, P = 0.13) and SCLC subgroup (RR 1.00, 95% CI 0.92–1.08, P = 0.16) (Fig. 6).

Fig. 6

Mortality subgroup analysis—NSCLC and SCLC forest plots

Incidence of thrombotic vascular events

A total of 5 included studies (I2 = 11%, P = 0.002) reported the occurrence of thrombotic vascular adverse events, including 218 (8.9%) of 2445 patients in the ESAs group and 123 (7.7%) of 1593 patients in the control group. The result showed that the incidence of adverse thrombovascular events was higher in patients using ESAs than in the control group (RR 1.40, 95% CI 1.13–1.72, P = 0.002) (Fig. 7).

Fig. 7

Forest plot of the incidence of thrombovascular adverse events between the erythropoietin and control groups

Blood transfusion requirements

A total of 7 included studies (I2 = 70%, P = 0.002) reported patients with transfusion requirements during treatment, including 496 (21.3%) of 2328 patients in the ESAs group and 543 (34.17%) of 1589 patients in the control group. Our studies showed that patients in the ESAs group had lower transfusion requirements than those in the control group (RR 0.56, 95% CI 0.44–0.72, P < 0.00001) (Fig. 8).

Fig. 8

Forest plot of blood transfusion requirements between the erythropoietin group and the control group

After sensitivity analysis of the 7 included studies by using Stata 14.0 statistical software, the results of the sensitivity analysis showed that the 2 included studies that published by Jürgen 2014 and Robert 2008 had a large impact on the results, and the study of Pere 2020 had the largest impact on the results (Fig. 9).

Fig. 9

Sensitivity analysis of blood transfusion requirements

A total of 6 studies were included, with 4 studies used Darbepoetin alfa and 2 studies used Epoetin alfa. And the results showed that patients treated with ESAs had lower blood transfusion requirements than control group in both the Darbepoetin alfa group (RR 0.57, 95% CI 0.41–0.79, P = 0.003) and the Epoetin alfa group (RR 0.68, 95% CI 0.47–0.99, P = 0.01) (Fig. 10).

Fig. 10

Subgroup analysis of transfusion requirements—forest plot using different types of erythropoietin

A total of 6 studies we included, of which 3 included NSCLC and 3 included SCLC, so we did subgroup analyses based on lung cancer type. And the results showed that in the NSCLC subgroup (RR 0.61, 95% CI 0.36–1.04, P = 0.009), there was no significant difference in the transfusion requirements between the ESAs group and the control group; whereas, the transfusion requirements in the ESAs group were lower (RR 0.51, 95% CI 0.40–0.65, P = 0.34) than those in the control group about the SCLC subgroup (Fig. 11).

Fig. 11

Subgroup analysis of blood transfusion requirements—NSCLC and SCLC forest plots

Incidence of adverse events

A total of 5 included studies (I2 = 47%, P = 0.11) reported that patients had experienced adverse events during treatment, including 1903 (81.8%) of 2326 patients in the ESAs group and 1205 (81.9%) of 1470 patients in the control group. The results showed that the difference of the two measures between the erythropoietin and control groups was not statistically significant (RR 0.98, 95% CI 0.95–1.00, P = 0.10) ( Fig. 12).

Fig. 12

Forest plot of the incidence of adverse events between the erythropoietin group and the control group