Scientific Papers

Metabolic/endocrine disorders in survivors of childhood-onset and cranial radiotherapy- treated ALL/NHL: a meta-analysis | Reproductive Biology and Endocrinology


Literature retrieval, characteristics of included studies and patients’ baseline demography

In the initial retrieval of the academic databases, 779 studies were scanned for their titles, and 594 studies were excluded as duplicates, study types that did not comply with the meta-analysis or obvious irrelevancy with the research objective of the present meta-analysis. The abstract of each retrieved article was carefully checked, and 170 studies were filtered out for different reasons, such as mechanistic studies based on basic research, studies without outcomes of interest, no access to full-text manuscripts, interventional studies instead of observational studies, or case studies. Finally, 15 studies [17,18,19,20,21,22,23,24,25,26,27,28,29,30,31] were included; a flowchart of study selection is shown in Fig. 1. Four single-arm studies were included for a meta-regression analysis to analyze the time-dependent effect of CRT treatment on different outcomes. The characteristics of the included studies and the patients’ demography at baseline are shown in Table 1, including author information, study year, study region, number of patients, study design, number of cohorts, hematologic neoplasm type, follow-up duration (years), male proportion (%), age at diagnosis of ALL/NHL (years), proportion of prepubertal stage at diagnosis (%), median age at the last visit, number of patients receiving CRT (%), and CRT doses. Clinical data from 4269 patients in the included studies were integrated and analyzed. In general, the follow-up duration in the included studies was approximately one decade. The ultimate quality check of the included clinical studies using NOS is shown in Table 1. Two independent investigators reached a consensus that the included studies were eligible for further meta-analyses.

Fig. 1
figure 1

Flow chart of literature retrieval and study inclusion

CRT in childhood influencing survivors’ height in the follow-up duration

Seven related studies were included in the analysis of adult height SDS (3724 patients) [17, 18, 22, 24, 27, 28, 30]. Figure 2 A shows that the adult height SDS was lower in CRT-treated patients (pooled SMD = -0.581, 95% CI: -0.649–-0.512). No heterogeneity was found in the extracted data (I-squared = 30.9%, p = 0.192), and the symmetric distribution of the included studies in the funnel plot (Fig. 2B) revealed no publication bias (p > 0.05). Moreover, sensitivity analysis, shown in Fig. 2C, suggests no inter-publication heterogeneity, and thus, no further subgroup analysis was required. Patients receiving CRT in childhood were likely to develop short stature (pooled OR = 2.289, 95% CI:1.674–3.130) in the follow-up period (Fig. 2D), and significant heterogeneity among the included studies was detected (I-squared = 79.5%, p = 0.001). The funnel plot in Fig. 2E shows no publication bias (P > 0.05). Six related studies were included in the analysis of the short stature (3716 patiens) [17,18,19, 21, 23, 24]. In patients with a history of CRT, the pooled prevalence rate of short stature was 0.103 (95% CI:0.09–0.116), as shown in Fig. 3A. Meta-regression was performed to analyze the correlation between short stature and different study years to test whether the proportion of patients with short stature in CRT-treated patients changed over time. The meta-regression coefficient between short stature and different study years was 0.931 (95% CI:0.703–1.233), indicating no time effect in CRT-treated patients who finally developed short stature (Fig. 3B). In addition, the GH secretion level, which was positively correlated with height, was assessed in 631 patients, and the pooled GH deficiency rate was 0.243 (95% CI:0.203–0.284) (Fig. 3C) [17, 18, 20, 21, 26,27,28,29,30]. The study year, which potentially reflects the state-of-the-art CRT technique, showed no relationship with GH deficiency prevalence by meta-regression (Fig. 3D), with a coefficient of 0.984 (95% CI:0.927–1.044). The importance of distinguishing GHD short stature from non-GHD short stature should also be considered. However, due to the lack of data on GH treatment in patients with GHD, this comparison could not be performed.

Fig. 2
figure 2

CRT in childhood influencing survivors’ height during follow-up. Value of height-SDS (A); funnel plot showing the symmetric distribution of included studies in this analysis (B); sensitivity analysis indicating no inter-publication heterogeneity (C); CRT treatment as risk factors in developing a short stature (D); publication bias analysis by funnel plot (E)

Fig. 3
figure 3

Pooled prevalence of short stature in included patients (A); meta-regression of time influence on short stature prevalence (B); pooled GH deficiency rate (C); meta-regression of time influence on the prevalence of GH deficiency (D)

Precocious puberty in ALL/NHL survivors who received CRT treatment in childhood

Five studies with 478 patiens were included in the analysis of precocious puberty [18, 19, 21, 26, 31]. CRT could increase the risk of precocious puberty (pooled OR = 2.937, 95% CI:1.281–6.736) in ALL/NHL survivors who received CRT treatment previously (Fig. 4A), and no heterogeneity was found in extracted data (I-squared < 1%, p = 0.439). In the funnel plot shown in Fig. 4B, the included studies were symmetrically distributed, indicating no publication bias (p > 0.05). In patients with previous CRT treatment, the pooled prevalence rate of precocious puberty was 11.8% (95% CI, 0.056–0.179; Fig. 4C). Meta-regression analysis (Fig. 4D) showed that the study year was not correlated with the prevalence of precocious puberty, with a coefficient of 0.982 (95% CI:0.889–1.084). Especially in female patients, the age at menarche tended to be earlier (pooled SMD = -0.235, 95% CI: -0.555–0.084) in patients with a history of CRT treatment (Fig. 5A), and no heterogeneity was found in extracted data (I-squared < 1%, p = 0.808). Additionally, the effects of CRT on sex hormone levels were evaluated. The testosterone level (nmol/L) showed no difference (pooled SMD = -0.065, 95% CI: -0.401 to 0.271) regardless of CRT treatment (Fig. 5B), and no heterogeneity was found in the extracted data (I-squared = 11.0%, p = 0.325). However, estradiol (ng/mL) was higher in patients previously treated with CRT (pooled SMD = 1.372, 95% CI:0.93–1.814), as shown in Fig. 5C.

Fig. 4
figure 4

Precocious puberty in ALL/NHL survivors who received CRT treatment in childhood: CRT increases the risk of precocious puberty (A); funnel plot showing no publication bias (B); pooled prevalence of precocious puberty (C); meta-regression of time influence on precocious puberty (D)

Fig. 5
figure 5

Difference in mean age at menarche (A); difference in testosterone level (B); difference in estradiol level (C)

Hypothyroidism and hypogonadism after CRT treatment

In analysis of 3907 patients, the risk of hypothyroidism was significantly increased (pooled OR = 2.057, 95% CI:1.510–2.801) in ALL/NHL survivors who had previously received CRT (Fig. 6A) [18,19,20, 24, 26, 28], and no heterogeneity was found in the extracted data (I-squared = 30.8%, p = 0.204). In the funnel plot shown in Fig. 6B, the included studies were symmetrically distributed, indicating no publication bias (p > 0.05). In patients with previous CRT treatment, the pooled prevalence rate of hypothyroidism was 7.1%, with a 95% CI ranging from 6.1 to 8.1% (Fig. 6C) [17,18,19,20, 24, 28, 30] after analyzing data from 3784 patients. Meta-regression (Fig. 6D) showed that the study year did not correlate with the prevalence of hypothyroidism with a coefficient of 0.983 (95% CI:0.931–1.039). However, the risk of hypogonadism was also significantly increased (pooled OR = 3.098, 95% CI:2.521–3.807) in ALL/NHL survivors who had previously received CRT (Fig. 7A) [18, 20, 24, 31], and no heterogeneity was found in the extracted data (I-squared = 41.9%, p = 0.16) from 3645 patients. In the funnel plot shown in Fig. 4F, the included studies are symmetrically distributed, indicating no publication bias (p > 0.05). In patients with previous CRT, the pooled prevalence rate of hypogonadism was 33.3%, with a 95% CI ranging from 31.4 to 35.2% (Fig. 7B). Meta-regression (Fig. 7C) [18, 20, 24, 31] showed that the study year did not correlate with the prevalence of hypogonadism with a coefficient of 0.953 (95% CI:0.827–1.098) in included 3645 patients. Distinguishing between TSH deficiency (CRT) and primary hypothyroidism (neck irradiation) is important. However, the influence of radiation therapy on hypothyroidism could not be compared because of the lack of relevant data.

Fig. 6
figure 6

Hypothyroidism and hypogonadism after CRT treatment: increased risk of hypothyroidism after CRT treatment (A); funnel plot showing a symmetric distribution of included studies in this analysis (B); pooled prevalence rate of hypothyroidism (C); meta-regression of time influence on hypothyroidism (D)

Fig. 7
figure 7

Increased risk of hypogonadism after CRT treatment (A); symmetrical distribution of included studies indicating no publication bias (B); pooled prevalence rate of hypogonadism (C); meta-regression of time influence on hypogonadism (D)

Other metabolic/endocrine problems in survivors after CRT in childhood

According to the abovementioned CRT-related short stature, the risk of vitamin D deficiency was higher (pooled OR = 3.575, 95% CI:1.226–10.427) in patients who had previously received CRT (Fig. 8A) [19, 22] from 110 patients, and no heterogeneity was found in the extracted data (I-squared < 0.1%, p = 0.634). The risk of overweight/obesity showed no difference (pooled OR = 1.278, 95% CI:0.675–2.421) between CRT-treated and non-CRT-treated patients (Fig. 8B) [18,19,20], and no heterogeneity was found in the extracted data (I-squared = 29.8%, p = 0.24) from 271 patients. In the funnel plot shown in Fig. 8C, the included studies are symmetrically distributed, indicating no publication bias (p > 0.05). By analyzing data reported from 349 patients with previous CRT treatment, the pooled prevalence rate of being overweight/obese was 19.2%, with a 95% CI ranging from 13.7 to 24.7% (Fig. 8D) [18,19,20, 29]. Meta-regression (Fig. 8E) showed that the study year did not correlate with being overweight/obese with a coefficient of 0.988 (95% CI:0.758–1.289).

Fig. 8
figure 8

Other metabolic/endocrine problems of survivors after CRT in childhood: risk of vitamin D deficiency after CRT treatment (A); risk of being overweight/obese after CRT treatment (B); funnel plot showing symmetric distribution of included studies in this analysis (C); pooled prevalence rate of being overweight/obese (D); meta-regression of time influence on being overweight/obese (E)



Source link