Body mass index and multiple myeloma: a systematic review and dose-response meta-analysis of prospective cohort studies

Abstract

Background

Findings on the association between body mass index (BMI) and risk of multiple myeloma (MM) are conflicting. Although some meta-analysis summarized available findings in this regard, some limitations, such as missing eligible articles and inclusion of ineligible articles, may have distorted their findings. Therefore, the current study was conducted to summarize available findings on the association between BMI and MM.

Methods

We searched online databases to identify prospective cohort studies that examined the association between BMI and risk of MM or its mortality.

Results

In total, 28 prospective cohort studies were included. During the follow-up periods ranging from 8.5 to 38 years, 18,923 cases of MM and 3,926 deaths from MM were recorded. The comparison between obese and normal-weight individuals revealed the summary relative risk (RR) of 1.17 (95% confidence interval (CI): 1.10–1.23) for MM incidence and 1.38 (95% CI: 1.25–1.52) for its mortality, indicating a significant positive association. Similar positive associations were also observed for overweight. Additionally, in the dose-response meta-analysis, each 5 kg/m² increase in BMI was associated with a 2% higher risk of MM (RR: 1.02, 95%CI: 1.01–1.03) and a 3% higher risk of MM mortality (RR: 1.03, 95% CI: 1.02–1.05). However, we found no significant association between underweight and the risk of MM (RR: 0.90, 95% CI: 0.67–1.21) and its mortality (RR: 0.94, 95% CI: 0.71–1.24).

Conclusions

In conclusion, we found that high BMI, defined as obesity and overweight, was associated with an increased risk of MM and its mortality. PROSPERO ID: CRD42023467626.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-026-16233-6.

Background

Multiple myeloma (MM) is a plasma cell malignancy that is increasing worldwide. According to the GBD 2021 study, global MM incidence and mortality more than doubled between 1990 and 2021, reaching approximately 148,755 new cases and 116,360 deaths in 2021 [1]. Notably, high BMI was identified as the primary modifiable risk factor, with its attributable burden increasing from 6.40% in 1990 to 7.96% in 2021 [2]. Since 2000, the five-year relative survival of this cancer has improved due to advances in treatment approaches; however, it is still low [3]. Therefore, the identification of modifiable risk factors might have a significant impact on reducing morbidity and mortality from this cancer.

Several risk factors, including family history of hematologic malignancy, chronic stimulation of the immune system, autoimmune disorders, exposure to ionizing radiation, and occupational exposure to pesticides or herbicides, have been identified for MM [4]. Recently, it has been shown that obesity is associated with increased risk of several malignancies, including esophageal, colon, breast, endometrial, and kidney cancers [5]. Several biological mechanisms may explain this relationship. Obesity is associated with increased inflammatory biomarkers, particularly interleukin-6 (IL-6), which plays an important role in plasma cell proliferation and differentiation [6, 7]. Additionally, obesity leads to insulin resistance and hyperinsulinemia, increasing insulin-like growth factor-I (IGF-I) concentrations, thereby stimulating myeloma cell proliferation and inhibiting apoptosis [8–10]. However, the associations of obesity with blood cancers such as MM have not been fully established because current evidence in this regard is conflicting [11–38]. Some studies reported that obesity was associated with an increased risk of MM [11, 16, 39], while others found no significant association [18, 22, 40, 41] and even inverse associations [42, 43]. In addition, some studies concluded that the association between obesity and MM is dose-dependent [25, 44].

Several previous meta-analyses have examined this association [45–48]; however, methodological limitations and the availability of new studies warrant an updated systematic review. Additionally, since the release of the last meta-analysis, several articles have been published on this association [11, 25, 28, 29, 36]. Given the above-mentioned points, our primary objective was to assess the association between BMI categories and MM risk and mortality, with dose-response analysis as a secondary aim. Therefore, we decided to summarize available findings on the association between BMI and risk of MM by performing an updated systematic review and dose-response meta-analysis on prospective cohort studies.

Methods

The current systematic review and meta-analysis study was conducted based on the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) protocol (Supplementary Table 1) [49].

Search strategy

To determine a wide range of relevant cohort analyses studying the correlation between BMI and the risk of MM and its mortality, we extensively examined online datasets, such as PubMed, Scopus, and ISI Web of Science, up to July 30, 2023. This comprehensive approach ensures the thoroughness of our research. Supplementary Table 2 displays the keywords utilized in the investigation procedure, designed to be inclusive and comprehensive. Additionally, we conducted an online search using Google Scholar, employing a combination of “body mass index” and “myeloma” keywords. The search engine results were arranged according to relevance to the keywords, and the first 500 articles were reviewed, ensuring an inclusive approach. No restrictions were imposed on publication time or language, further enhancing the inclusivity of our search. Furthermore, we examined the reference lists of the included articles to determine any potentially overlooked studies, leaving no potential source unexplored.

Inclusion criteria

We included prospective studies such as cohort, nested case-control, and case-cohort analyses that examined BMI as the exposure variable and its association with the risk of MM and its mortality as the outcome variable. Studies were required to include adult participants (≥ 18 years) with no prior diagnosis of MM or other hematologic malignancies at baseline.

Other inclusion standards were analyses that reported appropriate risk estimates containing odds ratio (OR), risk ratio (RR), hazard ratio (HR), or incidence rate ratio (IRR) with a 95% confidence interval (CI) for the relationship between BMI categories and the risk of MM, or supplied the essential data for estimating these effect sizes. Studies that provided raw data (number of cases and total participants in each BMI category) sufficient for calculating these effect sizes were also included. For studies with multiple publications from the same cohort, we selected the most recent publication with the largest number of cases or longest follow-up period to avoid duplication. Studies reporting results separately by sex or other subgroups were treated as independent studies in our analysis.

Exclusion criteria

Our current meta-analysis excluded letters, comments, reviews, meta-analyses, cross-sectional analyses, case-control analyses, ecological analyses, animal analyses, posters, abstracts, citations, and studies with insufficient data. Additionally, we excluded analyses conducted on patients with MM and those examining the relationship between childhood BMI and the risk of MM in adulthood.

Data extraction

Two researchers independently extracted the necessary data from each suitable analysis. The final decision was made by group consensus. From each article, we gathered the following information: first author’s name, year of publication, cohort name, analysis method, investigation area, participants’ age (mean or range), sex, model size, number of patients, duration of follow-up, techniques used to estimate BMI and MM, reported relative risk estimates for the relationship between BMI and MM/mortality, and the confounding variables modified in the statistical investigation. If an investigation documented multiple effect sizes for this relationship, we chose the one adjusted for the most confounding variables. Furthermore, we considered studies that reported results by sex or any other variables to be separate.

Quality assessment

The quality of investigations incorporated in the present meta-analysis was assessed using the Newcastle Ottawa Scale (NOS) [50]. This scale assigns 9 points to each investigation based on the subsequent parameters: 4 points for choosing contributors, 2 points for comparability, and 3 points for evaluating results. For the investigations incorporated in the present meta-analysis, the median score was 8. Investigations with a score of 8 or higher were regarded as high quality.

Statistical analysis

The meta-analysis included articles that utilized two types of BMI varieties: (1) according to the World Health Organization (WHO) cut-off points (< 18.5, 18.5–25, 25–30, and ≥ 30 kg/m²) and (2) according to the tertiles, quartiles, or quintiles of BMI. To conduct the meta-analysis, we used four methods: (1) the maximum versus lowest comparison, where we merged the RR of the maximum classification of BMI (≥ 30 kg/m²) or the maximum tertile, quartile, or quintile of BMI (for investigations that did not classify distributors according to WHO standards) in comparison with the lowest classification, (2) obesity vs. normal weight, where we assessed only the classification of BMI ≥ 30 kg/m² in comparison with normal-weight classification (18.5–25 kg/m²), (3) overweight vs. normal weight, where we evaluated only the classification of 25 ≤ BMI < 30 kg/m² in comparison with normal-weight classification (18.5–25 kg/m²), and (4) underweight vs. normal weight, where we evaluated only the classification of BMI < 18.5 kg/m² in comparison with normal-weight classification (18.5–25 kg/m²) [51].

Meta-analysis was performed using STATA software (version 14) and the metan command. To estimate the overview relative risks (RRs), we employed a fixed-effects model because there was low between-study heterogeneity in all the organizations evaluated in this meta-analysis [52]. Random effects models may create less trustworthy results because they give disproportionately more weight to smaller studies [53]. We assessed heterogeneity among the included investigations using the Q-statistic and I2 values. For the I2 statistic, < 25%, 25–50%, 50–75%, and > 75% indicated low, moderate, high, and extreme heterogeneity between studies, respectively. Subgroup studies were executed according to sample size, sex, location, follow-up duration, smoking adjustment, and study quality. To evaluate publication bias, we employed Egger’s regression asymmetry test [54]. Additionally, we employed the trim-and-fill approach to discover the impact of possible missing studies on the overall relative risk [55]. Moreover, a sensitivity analysis utilizing a fixed-effects model was executed to assess the reliance on the overall effect size of each study. In this study, each study was excluded to evaluate the effect of that study on the overall estimate.

For the linear dose-response analysis, we utilized the generalized least squares (glst command in STATA) estimation approach [56, 57]. First, we calculated the gradients specific to each examination and then merged these slopes to estimate an overall average slope. We employed a fixed-effects model to merge the study-specific slopes. The GLST approach required result dispersal, the total number of participants, and the effect sizes with the variance estimations for ≥ 3 quantitative exposure classifications. We utilized the median or mean BMI value for each classification in each investigation to estimate the related effect size. If an investigation documented a degree of BMI values, we calculated the midpoint for each classification. When the highest and lowest BMI classifications were not set, we supposed the intermission length to be the same as the adjoining intermission. Additionally, we investigated the probability of a non-linear dose-response relationship employing limited cubic splines with three tangles at the 10th, 50th, and 90th percentiles of the BMI distribution. We also considered the correlation between supplied risk assessments and integrated the study-specific assessments utilizing a linear mixed effects meta-analysis. This approach calculates the slopes specific to each investigation and then merges them to ascertain an overall average slope in a single phase. We tested the null hypothesis for non-linearity, considering the second spline’s coefficient to be zero. Statistical estimations were executed utilizing STATA version 14.0. We considered P < 0.05 statistically significant for all examinations, including the Cochran’s Q test.

Results

Literature search

In the initial search, 4,121 papers were found. We excluded duplicate papers (n = 400) and studies that did not meet our inclusion criteria based on screening titles and abstracts (n = 3,675). After these exclusions, 46 papers of potentially relevant studies were identified and included in the full-text assessment.

Of the 46 papers, we excluded 8 articles owing to assessing the link between BMI and MM survival (n = 2) [58, 59], having a case-control design (n = 2) [60, 61], being a review (n = 1) [62], and not reporting required data (n = 3) [63–65]. Also, we found 13 duplicate articles, of them, three were published on California Teachers Study (CTS) [28, 41, 66], 2 on Cancer Prevention Study-II (CPS-II) [28, 40], 2 on European Prospective Investigation into Cancer and Nutrition (EPIC) [38, 39], 2 on Metabolic Syndrome and Cancer Project (Me-Can) [20, 44], 3 on Swedish Construction Industry’s Organization for Working Environment (SCIOWE) [13, 42, 43], 2 on Japan Collaborative Cohort study (JACC) [32, 67]. We included only the study with higher quality, the greatest number of cases, or the longest follow-up period from each cohort [13, 20, 28, 32, 38]., and then excluded the duplicate articles [39–44, 66, 67].

Additionally, we found 8 articles on the same datasets examining different outcomes (incidence/mortality of MM) and included them in the meta-analysis [15–17, 26, 27, 33, 35, 36]. Some pooled analyses had partial overlap with individual studies; these were included in the main analysis with sensitivity analysis conducted to assess their impact [25, 28, 33–36]. Finally, 28 articles were included in the present systematic review and meta-analysis [11–38]; 20 articles investigated the association between BMI and the risk of MM [11–29, 38], and 8 papers assessed the link between BMI and mortality due to MM [30–37]. Figure 1 indicates a flow diagram of study selection.

Fig. 1.

Flow diagram of study selection for systematic review and meta-analysis of body mass index and multiple myeloma risk. CTS: California Teachers Study; CPS-II: Cancer Prevention Study-II; EPIC: European Prospective Investigation into Cancer and Nutrition; Me-Can: Metabolic Syndrome and Cancer Project; SCIOWE: Swedish Construction Industry’s Organization for Working Environment; JACC: Japan Collaborative Cohort Study; BMI: body mass index; MM: multiple myeloma

Characteristics of included studies

Table 1 shows the characteristics of the included studies. The studies were published between 1994 and 2022. The number of participants ranged from 522 to 5,243,978, with an age range between 13 and 104 years. In total, 22,914,751 participants were included in these studies. During the follow-up periods ranging from 8.5 to 38 years, 18,923 cases of MM and 3,926 deaths from MM were recorded. Of the 28 included articles, 4 papers included only men [13, 21, 23, 37], 3 papers included only women [12, 19, 27], and 6 papers reported risk estimates for men and women separately [14, 20, 26, 30, 31, 38]. Out of 28 articles, 13 papers were from United States (US) [12, 14–16, 23, 25–28, 30, 31, 35, 37], 9 articles were from non-US countries [11, 13, 17–19, 21, 22, 29, 32], and 6 papers were conducted on populations from different nations [20, 24, 33, 34, 36, 38]. To assess anthropometric measures, 15 papers used self-reported questionnaires [13, 15–17, 19, 22, 24–27, 30, 32, 34, 35, 37], 10 articles used trained staff-measured anthropometric data [11, 12, 14, 18, 20, 21, 23, 29, 31, 38], 2 articles used both self-reported questionnaires and measured anthropometric data [33, 36] and one article used both self-reported data and information from medical records [28]. MM was evaluated using medical records or cancer registries in 17 papers [11, 13–24, 26, 27, 29, 38], self-reported data in 2 articles [12, 25], and both medical records and self-reported data in one paper [28]. In the most included articles, relative risk estimates were controlled for some important confounders, such as age (n = 27), smoking status (n = 15), sex (n = 13), alcohol consumption (n = 11), and race (n = 11). The NOS scores of the included articles were between 6 and 9. We considered a median score of 8 for categorizing articles as high and low quality. Accordingly, 17 articles [11–14, 17–21, 23, 28–31, 33, 34, 36] had high quality, and others were considered low-quality studies (Supplementary Table 3).

Table 1.

Characteristics of included studies on the associations between body mass index and risk of multiple myeloma and its mortality¹

Author	Country/study name	Age (y2)	Sample size (n)	Sample size by sex (n)	Follow up (y3)	Case (n)	Exposure	Exposure assessment	Outcome	Outcome assessment	BMI (Median/cutoff point)	ES (95% CI)	Adjustment
Andreotti et al. 2010 [26]	US/AHS	> 18	67,947	M: 39,628	12	47	BMI	Self-reported	MM	Medical record	21.7	1.00	Race, meat consumption
											27.4	HR: 1.82 (0.48–2.89)
											32.4	HR: 1.35 (0.45–4.06)
				F: 28,319		21					21.7	1.00	Without any adjustment
											27.4	HR: 1.17 (0.43–3.12)
Bertrand et al. 2022 [28]	US/Pooled analysis of six US prospective cohorts (NHS, NHSII, HPFS, CPS-II, CTS, KPSC)	18->90	544,016	-	37	2,756	BMI	Self-reported/ medical record	MM	Self-reported/ medical record	16.2	HR: 0.83 (0.51–1.35)	Age, sex, cohort, race, education, smoking, alcohol, height, categorical early adult BMI
											20.7	1.00
											23.9	HR: 1.02 (0.88–1.17)
											27.4	HR: 1.01 (0.89–1.14)
											32.4	HR: 1.13 (0.97–1.32)
											37.4	HR: 1.06 (0.85–1.33)
											42.4	HR: 1.69 (1.23–2.31)
											Per 5 kg/m2 increase	HR: 1.09 (1.04–1.14)
Bhaskaran et al. 2014 [18]	UK/CPRD	≥ 16	5,243,978	-	25	2,969	BMI	Trained staff	MM	Medical record	15.2	RR: 1.01 (0.68–1.50)	Age, sex
											21.7	1.00
											27.4	RR: 1.05 (0.94–1.17)
											32.4	RR: 1.03 (0.89–1.20)
											37.4	RR: 1.20 (0.96–1.49)
											Per 5 kg/m2 increase	HR: 1.03 (0.98–1.09)	Age, diabetes, smoking, alcohol, socioeconomic status, calendar year
Blair et al. 2005 [27]	US/IWHS	55–69	37,083	F: 37,083	16	95	BMI	Self-reported	MM	Medical record	21.7	1.00	Age
											27.4	RR: 1.30 (0.78-2.00)
											32.5	RR: 1.50 (0.92–2.60)
Britton et al. 2008 [38]	Europe/EPIC	25–70	371,983	M: 141,425	8.5	139	BMI	Trained staff	MM	Medical record	22.4	1.00	Age, study center
											27.4	RR: 1.33 (0.79–2.23)
											32.4	RR: 1.17 (0.80–1.72)
				F: 230,558		129					22.4	1.00
											27.4	RR: 0.93 (0.55–1.56)
											32.4	RR: 1.06 (0.72–1.58)
Calle et al. 2003 [30]	US/CPS-II	≥ 30	900,053	M: 404,576	16	708	BMI	Self-reported	MM mortality	Medical record	21.7	1.00	Age, education, smoking, physical activity, alcohol, marital status, race, aspirin use, fat, and vegetable consumption
											27.4	RR: 1.18 (1.01–1.39)
											32.4	RR: 1.44 (1.10–1.89)
											37.4	RR: 1.71 (0.93–3.14)
				F: 495,477		620					21.7	1.00	Age, education, smoking, physical activity, alcohol, marital status, race, aspirin use, estrogen-replacement therapy, fat, and vegetable consumption
											27.4	RR: 1.12 (0.93–1.34)
											32.4	RR: 1.47 (1.13–1.91)
											37.4	RR: 1.44 (0.91–2.28)
Cao et al. 2020 [11]	UK/UK Biobank	37–73	390,575	-	10	370	BMI	Trained staff	MM	Medical record	21.7	1.00	Sex, age, race, Townsend deprivation index, qualification, employment status, alcohol, smoking
											27.5	HR: 0.97 (0.76–1.26)
											32.5	HR: 1.32 (1.01–1.74)
Chiu et al. 2006 [31]	US/CHADP	40	35,420	M: 20,314	35	38	BMI	Study nurses	MM mortality	Medical record	22.7	1.00	Age, education, smoking, race, post-load glucose level
											25.3	HR: 0.99 (0.35–2.83)
											27.4	HR: 1.61 (0.63–4.13)
											30.5	HR: 1.62 (0.63–4.19)
				F: 15,106		28					19.7	1.00
											22.1	HR: 1.69 (0.42–6.80)
											24.5	HR: 2.06 (0.54–7.83)
											28.9	HR: 1.97 (0.51–7.53)
De roos et al. 2010 [12]	US/ WHIOS	50–79	81,219	F: 81,219	14	91	BMI	Certified staff	MM	Self-reported	22.4	1.00	Age, race, education, US region, smoking
											27.4	HR: 1.03 (0.65–1.63)
											32.4	HR: 0.66 (0.33–1.33)
											37.4	HR: 0.83 (0.37–1.87)
Fernberg et al. 2007 [13]	Sweden/SCIOWE	14–82	336,381	M: 336,381	33.5	519	BMI	Self-reported	MM	Medical record	21.7	1.00	Age, tobacco use
											27.5	IRR: 1.04 (0.86–1.24)
											32.5	IRR: 0.70 (0.46–1.06)
Friedman et al. 1994 [14]	US/KPMCP	13–92	143,574	-	21	White men:66	BMI	Trained staff	MM	Medical record	22.1	1.00	Age
											24.1	RR: 1.20 (0.50–2.90)
											26.2	RR: 1.60 (0.70–3.70)
											28.4	RR: 1.70 (0.70–3.80)
						Black men:34					22.1	1.00
											24.1	RR: 3.40 (0.70–16.00)
											26.2	RR: 2.90 (0.60–13.10)
											28.4	RR: 2.80 (0.60–12.10)
						White women:49					19.9	1.00
											21.9	RR: 0.50 (0.20–1.30)
											24.3	RR: 0.80 (0.30–1.70)
											27.1	RR: 0.70 (0.30–1.60)
						Black women:18					21.9	1.00
											24.3	RR: 0.60 (0.10–2.80)
											27.1	RR: 0.90 (0.20–3.20)
Gray et al. 2011 [37]	US/HAHS	46.1	19,209	M: 19,209	36	58	BMI	Self-reported	MM mortality	Medical record	22.0	1.00	Age
											23.5	HR: 0.66 (0.27–1.61)
											25.0	HR: 1.81 (0.90–3.66)
											26.5	HR: 1.34 (0.63–2.83)
											Per 2.55 kg/m2 increase	HR: 1.15 (0.90–1.48)
Hofmann et al. 2012 [15]	US/PLCO Cancer Screening Trial	55–74	522	-	17	174	BMI	Self-reported	MM	Medical record	Q4 vs. Q1	OR: 1.39 (0.83–2.35)	Age, sex, race, date and time of day of phlebotomy, study year of specimen collection
											Per 5 kg/m2 increase	OR: 1.14 (0.94–1.39)
Hofmann et al. 2013 [16]	US/NIH-AARP Diet and Health Study	50–71	485,049	-	11	476	BMI	Self-reported	MM	Medical record	16.5	HR: 0.30 (0.04–2.17)	Age, sex, race
											20.4	1.00
											23.7	HR: 1.02 (0.73–1.43)
											27.4	HR: 1.09 (0.80–1.48)
											32.4	HR: 1.26 (0.89–1.78)
											37.4	HR: 1.55 (1.01–2.39)
											Per 5 kg/m2 increase	HR: 1.10 (1.00-1.22)
Kanda et al. 2010 [17]	Japan/JPHCPS	40–69	94,547	-	16	88	BMI	Self-reported	MM	Medical record	16.2	HR: 0.56 (0.13–2.36)	Age, sex, study area, smoking, alcohol
											20.7	HR: 0.70 (0.42–1.15)
											23.9	1.00
											27.4	HR: 0.79 (0.45–1.38)
											32.4	HR: 0.76 (0.18–3.20)
											Per 1 unit increase	HR: 1.01 (0.95–1.09)
Khan et al. 2006 [32]	Japan/JACC	40–79	109,698	-	15	98	BMI	Self-reported	MM mortality	Medical record	15.1	HR: 0.79 (0.32–1.98)	Age, sex
											21.7	1.00
											27.5	HR: 0.72 (0.39–1.33)
											32.5	HR: 2.79 (1.01–7.69)
Marinac et al. 2018 [25]	US/Pooled analysis of three US prospective cohort (HPFS, NHS, WHS)	≥ 30	202,634	-	36	471	BMI	Self-reported	MM	Self-reported	20.7	1.00	Age, race, and change in BMI since young adulthood
											23.9	HR: 1.20 (0.75–1.91)
											26.2	HR: 1.15 (0.77–1.71)
											28.7	HR: 1.55 (1.11–2.17)
											31.2	HR: 1.48 (0.75–2.90)
											Per 5 kg/m2 increase	HR: 1.20 (1.06–1.37)
Murphy et al. 2013 [19]	UK/MWS	50–65	1,249,897	F: 1,249,897	13	1,503	BMI	Self-reported	MM	Medical record	23.9	1.00	Height, alcohol, smoking, socioeconomic status
											28.6	RR: 1.09 (0.98–1.23)
											34.5	RR: 1.17 (1.02–1.35)
											Per 10 kg/m2 increase	RR: 1.17 (1.02–1.33)
Nagel et al. 2012 [20]	Austria, Norway, and Sweden/Me-Can project	< 29->70	575,386	M: 288,068	> 10	301	BMI	Trained staff	MM	Medical record	Q1	1.00	Age, smoking
											Q2	RR: 1.00 (0.63–1.60)
											Q3	RR: 1.58 (1.03–2.41)
											Q4	RR: 1.64 (1.07–2.49)
											Q5	RR: 1.27 (0.81–1.98)
				F: 287,318		196					Q1	1.00
											Q2	RR: 0.85 (0.48–1.50)
											Q3	RR: 0.84 (0.48–1.47)
											Q4	RR: 0.98 (0.57–1.66)
											Q5	RR: 1.17 (0.70–1.97)
Oh et al. 2005 [21]	Korea/KNHIC	≥ 20	781,283	M: 781,283	10	103	BMI	Trained staff	MM	Medical record	16.2	RR: 1.19 (0.29–4.96)	Age, smoking, alcohol, physical activity, FH of cancer, residency area
											20.7	1.00
											23.9	RR: 1.72 (1.11–2.68)
											25.9	RR: 0.96 (0.51–1.77)
											28.4	RR: 0.98 (0.30–3.32)
Parr et al. 2010 [34]	Asia, Australia, and New Zealand/ APCSC	≥ 20	401,215	-	38	69	BMI	Self-reported	MM mortality	Medical record	17.5	HR: 1.94 (0.57–6.68)	Age, smoking
											22.0	1.00
											26.7	HR: 0.87 (0.54–1.41)
											32.9	HR: 1.20 (0.59–2.43)
											Per 5 kg/m2 increase	HR: 1.05 (0.73–1.50)
Pylypchuk et al. 2009 [22]	Netherlands/ NCSDC	55–69	4,774	-	13.3	279	BMI	Self-reported	MM	Medical record	21.7	1.00	Age, sex
											27.4	HR: 1.23 (0.95–1.58)
											32.4	HR: 1.13 (0.68–1.88)
											Per 4 kg/m2 increase	HR: 1.13 (0.97–1.31)
Recalde et al. 2021 [29]	Spain	≥ 18	3,658,417	-	12	3,270	BMI	Trained staff	MM	Medical record	T3 vs. T1	HR: 1.20 (1.08–1.35)	Smoking, alcohol, nationality, the MEDEA deprivation index, diabetes, sex, age
											Per 5 kg/m2 increase	HR: 1.09 (1.04–1.15)
Samanic et al. 2004 [23]	US/Male US veterans	18–100	4,500,700	White men: 3,668,486	27	3,021	BMI	Trained staff	MM	Medical record	≥ 30 vs. <30	RR: 1.22 (1.05–1.40)	Age, calendar year
				Black men: 832,214		1,598					≥ 30 vs. <30	RR: 1.26 (1.02–1.56)
Soderberg et al. 2009 [24]	Sweden and Finland/Study of Swedish and Finnish twins	18–96	70,067	-	28	140	BMI	Self-reported	MM	Medical record	21.7	1.00	Age, sex, country
											27.4	RR: 1.00 (0.70–1.50)
											32.4	RR: 2.10 (1.10–3.70)
Sonderman et al. 2016 [35]	US/Pooled analysis of 7 cohorts (NIH-AARP, BWHS, CPSII, MEC, PLOC, SCCS, AHS2	20–104	239,597	-	26.5	491	BMI	Self-reported	MM mortality	Medical record	16.7	HR: 1.18 (0.48–2.88)	Age, sex
											21.7	1.00
											27.4	HR: 1.07 (0.86–1.33)
											32.4	HR: 1.14 (0.87–1.49)
											47.4	HR: 1.43 (1.03–1.97)
											Per 5 kg/m2 increase	HR: 1.08 (0.99–1.17)
Teras et al. 2014 [33]	US, Sweden, Australia, and Norway/20 cohorts of the National Cancer Institute Cohort Consortium	19–83	1,564,218	-	> 10	1,388	BMI	Self-reported/ trained staff	MM mortality	Medical record	16.7	HR: 1.21 (0.71–2.06)	Race, education, sex, marital status, alcohol, physical activity, smoking
											19.7	HR: 1.02 (0.79–1.32)
											21.9	1.00
											23.9	HR: 1.22 (1.01–1.47)
											26.2	HR: 1.15 (0.95–1.38)
											28.7	HR: 1.24 (1.01–1.52)
											32.4	HR: 1.23 (0.99–1.52)
											37.4	HR: 1.52 (1.15–2.02)
											Per 5 kg/m2 increase	HR: 1.09 (1.03–1.16)
Ugai et al. 2019 [36]	Japan, China, Taiwan, Korea, India, and Singapore/Asian Cohort Consortium	54	805,309	-	> 10	428	BMI	Self-reported/ trained staff	MM mortality	Medical record	15.2	HR: 0.80 (0.52–1.24)	Age, sex, smoking, alcohol, education
											21.7	1.00
											27.4	HR: 1.17 (0.94–1.47)
											32.4	HR: 1.61 (0.99–2.64)
											Per 1 kg/m2 increase	HR: 1.04 (1.01–1.07)

¹Abbreviation: y year, n number, ES effect size, CI confidence interval, US United States, AHS Agricultural Health Study, M Male, F Female, BMI body mass index, MM multiple myeloma, HR hazard ratio, NHS Nurses’ Health Study, NHSII Nurses’ Health Study II, HPFS Health Professionals Follow-up Study, CPS-II Cancer Prevention Study-II, CTS California Teachers Study, KPSC Kaiser Permanente Southern California, kg/m² Kilograms/meter², UK United Kingdom, CPRD Clinical Practice Research Datalink, IWHS Iowa Women’s Health Study, RR risk ratio, EPIC European Prospective Investigation into Cancer and Nutrition, CHADP Chicago Heart Association Detection Project, WHIOS Women’s Health Initiative Observational Study, SCIOWE Swedish Construction Industry’s Organization for Working Environment, IRR incidence rate ratio, KPMCP Kaiser Permanente Medical Care Program, HAHS Harvard Alumni Health Study, PLCO Cancer Screening Trial Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial, Q quartile quintile, OR odd ratio, NIH-AARP National Institutes of Health-American Association of Retired Persons, JPHCPS Japan Public Health Center-based Prospective Study, JACC Japan Collaborative Cohort study, WHS Women’s Health Study, MWS Million Women Study, Me-Can project metabolic syndrome and cancer project, KNHIC Korea National Health Insurance Corporation Study, FH family history, APCSC Asia-Pacific Cohort Studies Collaboration, NCSDC Netherlands Cohort Study on Diet and Cancer, MEDEA Mortalidad en áreas pequeñas Españolas y Desigualdades Socioeconómicas y Ambientales, T tertile, BWHS Black Women’s Health Study, MEC Multiethnic Cohort Study, SCCS Southern Community Cohort Study, AHS2 Adventist health study 2

²Presented as mean or range

³Presented as the maximum duration of follow-up

Findings from the systematic review

Out of 20 articles on the link between BMI and the risk of MM, 8 papers reported a positive association [11, 16, 19, 23–25, 28, 29], and others reported no significant association. For MM mortality, 4 studies reported a positive association with BMI [32, 33, 35, 36], and other studies showed no significant association between BMI and MM mortality.

Meta-analysis on BMI and risk of MM

Highest vs. lowest comparison of BMI

Twenty-one studies from 18 publications evaluated the association between BMI and the risk of MM [11–13, 15–20, 22–29, 38]. Among 17,886,856 participants included in these studies, 18,632 cases of MM were recorded. Comparing the highest categories of BMI with the lowest categories, the summary RR for the risk of MM was 1.18 (95% CI: 1.12–1.24, I²= 0.0%, P = 0.47), indicating a significant positive association (Table 2).

Table 2.

Summary risk estimates for the association between body mass index and the risk of multiple myeloma Incidence¹

	No of effect sizes	Pooled relative risk (95% CI)2	P	I2 (%)3	P-heterogeneity4
The highest versus lowest comparison
Highest versus lowest
Overall	21	1.18 (1.12–1.24)	< 0.001	0.0	0.47
Subgroup analysis
Sample size, participants
≥ 10,000	19	1.18 (1.12–1.24)	< 0.001	6.9	0.37
< 10,000	2	1.25 (0.87–1.80)	0.23	0.0	0.58
Sex
Men and women	10	1.18 (1.11–1.26)	< 0.001	0.0	0.47
Men	6	1.19 (1.07–1.32)	0.002	25.8	0.24
Women	5	1.15 (1.02–1.29)	0.03	2.5	0.39
Study location
US	9	1.21 (1.12–1.31)	< 0.001	0.0	0.63
Non-US countries	12	1.15 (1.08–1.23)	< 0.001	12.5	0.32
Adjustment for smoking
Yes	10	1.15 (1.08–1.21)	< 0.001	22.0	0.24
No	11	1.26 (1.15–1.39)	< 0.001	0.0	0.88
Study quality5
High quality	12	1.16 (1.10–1.22)	< 0.001	13.9	0.31
Low quality	9	1.32 (1.13–1.53)	< 0.001	0.0	0.79
Obesity versus normal
Overall	17	1.17 (1.10–1.23)	< 0.001	16.0	0.27
Subgroup analysis
Sample size, participants
≥ 10,000	16	1.17 (1.10–1.24)	< 0.001	21.2	0.21
< 10,000	1	1.13 (0.68–1.88)	0.64	-	-
Sex
Men and women	8	1.17 (1.08–1.26)	< 0.001	14.3	0.32
Men	5	1.18 (1.06–1.32)	0.003	39.8	0.16
Women	4	1.14 (1.01–1.30)	0.03	26.8	0.25
Study location
US	8	1.21 (1.12–1.31)	< 0.001	0.0	0.56
Non-US countries	9	1.12 (1.04–1.22)	0.003	30.6	0.17
Adjustment for smoking
Yes	7	1.12 (1.05–1.20)	0.001	42.0	0.11
No	10	1.26 (1.15–1.39)	< 0.001	0.0	0.83
Study quality5
High quality	9	1.15 (1.08–1.22)	< 0.001	34.2	0.14
Low quality	8	1.31 (1.12–1.53)	0.001	0.0	0.71
Overweight versus normal
Overall	21	1.07 (1.02–1.13)	0.009	0.0	0.65
Subgroup analysis
Sample size, participants
≥ 10,000	20	1.07 (1.01–1.13)	0.02	0.0	0.66
< 10,000	1	1.23 (0.95–1.59)	0.11	-	-
Sex
Men and women	8	1.06 (0.99–1.13)	0.10	5.7	0.39
Men	6	1.13 (0.97–1.32)	0.12	27.9	0.23
Women	7	1.08 (0.98–1.20)	0.14	0.0	0.91
Study location
US	11	1.10 (1.00-1.21)	0.05	19.2	0.26
Non-US countries	10	1.06 (1.00-1.13)	0.07	0.0	0.89
Adjustment for smoking
Yes	8	1.04 (0.98–1.10)	0.19	0.0	0.95
No	13	1.22 (1.08–1.38)	0.001	0.0	0.65
Study quality5
High quality	12	1.05 (0.99–1.11)	0.13	0.0	0.65
Low quality	9	1.21 (1.06–1.37)	0.003	0.0	0.83
Underweight versus normal
Overall	5	0.90 (0.67–1.21)	0.49	0.0	0.71
Subgroup analysis
Sex
Men and women	4	0.89 (0.66–1.20)	0.45	0.0	0.57
Men	1	1.19 (0.29–4.92)	0.81	-	-
Study location
US	2	0.78 (0.49–1.26)	0.31	0.0	0.33
Non-US countries	3	0.98 (0.68–1.42)	0.93	0.0	0.72
Adjustment for smoking
Yes	4	0.92 (0.69–1.24)	0.60	0.0	0.81
No	1	0.30 (0.04–2.21)	0.24	-	-
Study quality5
High quality	4	0.92 (0.69–1.24)	0.60	0.0	0.81
Low quality	1	0.30 (0.04–2.21)	0.24	-	-
Linear dose-response association
BMI (per 5 kg/m2 increase):
Overall	23	1.02 (1.01–1.03)	< 0.001	46.8	0.007
Subgroup analysis
Sample size, participants
≥ 10,000	21	1.02 (1.01–1.03)	< 0.001	49.9	0.005
< 10,000	2	1.03 (0.99–1.08)	0.11	0.0	0.32
Sex
Men and women	10	1.05 (1.03–1.07)	< 0.001	49.8	0.04
Men	6	1.01 (0.98–1.03)	0.48	28.7	0.22
Women	7	1.01 (1.00-1.02)	0.21	0.0	0.62
Study location
US	12	1.06 (1.03–1.09)	< 0.001	34.9	0.11
Non-US countries	11	1.01 (1.00-1.03)	0.008	34.2	0.12
Adjustment for smoking
Yes	9	1.02 (1.01–1.03)	0.003	67.2	0.002
No	14	1.04 (1.01–1.06)	0.002	12.4	0.32
Study quality5
High quality	13	1.02 (1.01–1.03)	0.003	58.8	0.004
Low quality	10	1.04 (1.01–1.06)	0.002	7.3	0.37

¹Abbreviation: BMI body mass index, CI confidence interval, US United States

²Obtained from the fixed-effects model

³Inconsistency- the percentage of variation across studies due to heterogeneity

⁴Obtained from the Q-test

⁵Studies with a median score of 8 or more, based on the NOS, were considered a high-quality study

Obese (BMI ≥ 30) vs. normal-weight (BMI: 18.5–24.9) individuals

Fifteen papers (17 studies) were included in this analysis [11–13, 16–19, 22–28, 38]. These studies included 13,652,531 participants and 14,691 cases of MM. The summary effect size for the risk of MM, comparing obese and normal-weight participants, was 1.17 (95% CI: 1.10–1.23, P < 0.001), indicating a significant positive association between obesity and the risk of MM (Table 2). Among these studies, no significant heterogeneity was found (I²= 16.0%, P = 0.27).

Overweight (BMI: 25-29.9) vs. normal-weight (BMI: 18.5–24.9) individuals

Twenty-one studies from 16 articles investigated the association between overweight and MM risk [11–14, 16–19, 21, 22, 24–28, 38]. These studies contained 10,105,007 participants and 10,363 cases of MM. When overweight is compared with normal-weight participants, a significant positive association was found between overweight and the risk of MM (RR: 1.07, 95% CI: 1.02–1.13, I²= 0.0%; P = 0.65) (Table 2).

Underweight (BMI < 18.5) vs. normal-weight (BMI: 18.5–24.9) individuals

Five articles with a total sample size of 7,148,873 participants and 6,392 cases of MM assessed the link between underweight and the risk of MM [16–18, 21, 28]. Combining the results from these articles, we found no significant association between underweight and MM risk (RR: 0.90, 95%CI: 0.67–1.21, I² = 0.0; P = 0.71) (Table 2).

Dose-response association

Twenty-three studies from 18 papers were eligible for the linear dose-response analysis of BMI and the risk of MM [11–19, 21, 22, 24–29, 38]. In this analysis, we found a significant positive association between a 5 kg/m² increase in BMI and MM risk (RR: 1.02, 95%CI: 1.01–1.03, I² = 46.8; P = 0.007) (Table 2). Sixteen articles were included in the non-linear dose-response analysis [11–14, 16–19, 21, 22, 24–28, 38]. A non-linear association was found between BMI and the risk of MM (P = 0.01 for non-linearity), in which the positive association between BMI and MM risk intensified from 27 kg/m² to the highest BMI (Fig. 2).

Fig. 2.

Non-linear dose-response association of body mass index (based on kg/m²) with risk of multiple myeloma. The solid lines indicate the spline model. The dashed lines present the 95% CI. BMI: body mass index, RR: relative risk, MM: multiple myeloma

Meta-analysis on BMI and risk of MM mortality

Highest vs. lowest comparison of BMI

Eight studies from 7 articles, with a total sample size of 4,040,404 participants and 3,840 deaths from MM, were included [30–36]. Comparing the highest categories of BMI with the lowest categories, the summary RR for the risk of mortality from MM was 1.38 (95% CI: 1.25–1.52, I²= 0.0%, P = 0.77), indicating a significant positive association (Table 3).

Table 3.

Summary risk estimates for the association between body mass index and the risk of multiple myeloma mortality¹

	No of effect sizes	Pooled relative risk (95% CI)2	P	I2 (%)3	P-heterogeneity4
The highest versus lowest comparison
Highest versus lowest
Overall	8	1.38 (1.25–1.52)	< 0.001	0.0	0.77
Subgroup analysis
Sex
Men and women	5	1.33 (1.17–1.50)	< 0.001	0.0	0.55
Men	2	1.49 (1.17–1.89)	0.001	0.0	0.86
Women	1	1.46 (1.16–1.84)	0.001	-	-
Study location
US	4	1.38 (1.21–1.57)	< 0.001	0.0	0.67
Non-US countries	3	1.60 (1.10–2.32)	0.01	0.0	0.41
Both	1	1.33 (1.12–1.58)	0.001	-	-
Adjustment for smoking
Yes	6	1.41 (1.26–1.58)	< 0.001	0.0	0.94
No	2	1.29 (1.05–1.58)	0.01	56.7	0.13
Study quality5
High quality	6	1.41 (1.26–1.58)	< 0.001	0.0	0.94
Low quality	2	1.29 (1.05–1.58)	0.01	56.7	0.13
Obesity versus normal
Overall	8	1.38 (1.25–1.52)	< 0.001	0.0	0.77
Subgroup analysis
Sex
Men and women	5	1.33 (1.17–1.50)	< 0.001	0.0	0.55
Men	2	1.49 (1.17–1.89)	0.001	0.0	0.86
Women	1	1.46 (1.16–1.84)	0.001	-	-
Study location
US	4	1.38 (1.21–1.57)	< 0.001	0.0	0.67
Non-US countries	3	1.60 (1.10–2.32)	0.01	0.0	0.41
Both	1	1.33 (1.12–1.58)	0.001	-	-
Adjustment for smoking
Yes	6	1.41 (1.26–1.58)	< 0.001	0.0	0.94
No	2	1.29 (1.05–1.58)	0.01	56.7	0.13
Study quality5
High quality	6	1.41 (1.26–1.58)	< 0.001	0.0	0.94
Low quality	2	1.29 (1.05–1.58)	0.01	56.7	0.13
Overweight versus normal
Overall	10	1.14 (1.06–1.23)	0.001	0.0	0.85
Subgroup analysis
Sex
Men and women	5	1.13 (1.02–1.25)	0.02	2.2	0.39
Men	3	1.19 (1.02–1.39)	0.02	0.0	0.92
Women	2	1.11 (0.94–1.31)	0.23	0.0	0.77
Study location
US	6	1.13 (1.03–1.25)	0.01	0.0	0.96
Non-US countries	3	1.06 (0.88–1.29)	0.53	31.7	0.23
Both	1	1.19 (1.04–1.37)	0.01	-	-
Adjustment for smoking
Yes	7	1.16 (1.07–1.25)	< 0.001	0.0	0.91
No	3	1.04 (0.86–1.27)	0.68	0.0	0.39
Study quality5
High quality	7	1.16 (1.07–1.25)	< 0.001	0.0	0.91
Low quality	3	1.04 (0.86–1.27)	0.68	0.0	0.39
Underweight versus normal
Overall	6	0.94 (0.71–1.24)	0.65	0.0	0.49
Subgroup analysis
Sex
Men and women	5	0.99 (0.74–1.32)	0.93	0.0	0.56
Women	1	0.53 (0.20–1.40)	0.20	-	-
Study location
US	2	0.82 (0.42–1.58)	0.55	28.9	0.24
Non-US countries	3	0.87 (0.60–1.26)	0.45	0.0	0.40
Both	1	1.21 (0.71–2.06)	0.48	-	-
Adjustment for smoking
Yes	4	0.93 (0.68–1.27)	0.65	26.0	0.26
No	2	0.97 (0.51–1.84)	0.92	0.0	0.54
Study quality5
High quality	4	0.93 (0.68–1.27)	0.65	26.0	0.26
Low quality	2	0.97 (0.51–1.84)	0.92	0.0	0.54
Linear dose-response association
BMI (per 5 kg/m2 increase):
Overall	10	1.03 (1.02–1.05)	< 0.001	0.0	0.81
Subgroup analysis
Sex
Men and women	5	1.03 (1.01–1.05)	0.004	1.4	0.40
Men	3	1.03 (1.01–1.06)	0.008	0.0	0.75
Women	2	1.03 (1.00-1.07)	0.07	0.0	0.43
Study location
US	6	1.03 (1.01–1.05)	0.001	0.0	0.89
Non-US countries	3	1.03 (0.99–1.07)	0.14	0.0	0.93
Both	1	1.09 (1.03–1.16)	0.004	-	-
Adjustment for smoking
Yes	7	1.04 (1.02–1.05)	< 0.001	0.0	0.68
No	3	1.02 (0.99–1.05)	0.13	0.0	0.75
Study quality5
High quality	7	1.04 (1.02–1.05)	< 0.001	0.0	0.68
Low quality	3	1.02 (0.99–1.05)	0.13	0.0	0.75

¹Abbreviation: BMI body mass index, CI confidence interval, US United States

²Obtained from the fixed-effects model

³Inconsistency- the percentage of variation across studies due to heterogeneity

⁴Obtained from the Q-test

⁵Studies with a median score of 8 or more, based on the NOS, were considered a high-quality study

Obese (BMI ≥ 30) vs. normal-weight (BMI: 18.5–24.9) individuals

Seven papers (8 studies) compared obese with normal-weight individuals in terms of the risk of MM death [30–36]. These studies included a total of 4,040,404 participants and 3,840 deaths from MM. Meta-analysis of this comparison revealed a significant positive association between obesity and risk of MM death (RR: 1.38, 95% CI: 1.25–1.52, I²= 0.0%, P = 0.77) (Table 3).

Overweight (BMI: 25-29.9) vs. normal-weight (BMI: 18.5–24.9) individuals

Ten studies from 8 articles assessed the association between overweight and death from MM [30–37]. Among 4,074,719 participants included in these studies, 3,926 deaths due to MM were recorded. Combining the results from these articles, we found a significant positive association between overweight and MM mortality (RR:1.14, 95% CI: 1.06–1.23, I²= 0.0%, P = 0.85) (Table 3).

Underweight (BMI < 18.5) vs. normal-weight (BMI: 18.5–24.9) individuals

Six studies with a total of 3,135,143 participants and 2,502 deaths from MM were included in this Sects [31–36]. The summary effect size for the risk of MM mortality, comparing underweight and normal-weight participants, was 0.94 (95% CI: 0.71–1.24, I²= 0.0, P = 0.49), indicating no significant association between underweight and the risk of MM mortality (Table 3).

Dose-response association

Ten studies from 8 papers were included in the linear dose-response analysis [30–37]. A significant positive association was found between a 5 kg/m² increase in BMI and the risk of MM mortality (RR: 1.03, 95% CI: 1.02–1.05, I²= 0.0, P = 0.81) (Table 3). Eight papers had required data for inclusion in the non-linear dose-response analysis [30–37]. In this analysis, we found a non-linear association between BMI and risk of MM mortality (P = 0.01 for non-linearity), in which a significant positive association was seen from 19 kg/m² to the highest BMI; however, the risk increase was stronger at the BMI of ≥ 27 kg/m² (Fig. 3).

Fig. 3.

Non-linear dose-response association of body mass index (based on kg/m²) with risk of multiple myeloma mortality. The solid lines indicate the spline model. The dashed lines present the 95% CI. BMI: body mass index, RR: relative risk, MM: multiple myeloma

Subgroup and sensitivity analyses, and publication bias

Tables 2 and 3 show findings from the subgroup analysis of BMI and the risk of MM and its mortality. According to this analysis, a significant positive association was found between BMI (highest versus lowest comparison and obesity versus normal-weight comparison) and risk of MM in all subgroups except those studies with a low sample size (< 10,000 participants). The same findings were also observed for the comparison between overweight and normal-weight; however, the significant positive association between overweight and MM became non-significant in studies that controlled for smoking and those with high quality. For comparison between underweight and normal-weight, we found no significant association in any subgroups (Table 2). Similar findings were also seen for MM mortality, in which its positive association with overweight and obesity was significant among high-quality studies (Table 3).

Findings from the sensitivity analysis indicated that the exclusion of pooled analyses, conducted by Sonderman et al. 2016 [35], Teras et al. 2014 [33], and Ugai et al. 2019 [36], did not change our findings on the association between BMI and MM mortality (Supplementary Tables 4 & Supplementary Fig. 1). Sensitivity analysis for the other associations showed that the exclusion of any single study from the analysis did not alter the overall effect sizes. By using Egger’s linear regression test, we found no substantial publication bias in all associations (P > 0.10).

Discussion

In the current study, our primary findings showed that high BMI, defined as obesity (BMI ≥ 30 kg/m²) and overweight (BMI between 25 and 30 kg/m²), was associated with an increased risk of MM and its mortality. In addition, our secondary dose-response analysis revealed that each 5 kg/m² increase in BMI was associated with a 2% higher risk of MM and a 3% higher risk of MM mortality. Despite the statistical significance of this association, the small effect size suggests that BMI may not be a major risk factor for MM, or that its independent effect is relatively limited. In contrast, no significant association was found between underweight (BMI < 18.5 kg/m²) and MM incidence/mortality.

MM is one of the blood cancers with a poor prognosis [68]. Several risk factors are involved in the etiology of MM [4]. Among modifiable risk factors, high attention has been paid to obesity. However, current evidence in this regard is conflicting [11, 16, 18, 22, 39–43]. Our findings were in agreement with previous meta-analyses that reported a significant positive association between obesity indicators and MM incidence/mortality [45–48]. Recent comprehensive evidence from umbrella reviews further supports these associations. Chen et al. (2023) in an umbrella review of 18 meta-analyses examining BMI and various cancer risks found that overweight and obesity enhance the incidence of multiple myeloma, with each 5 kg/m² increase in BMI associated with a 12% higher risk [69]. Similarly, Sun et al., in a large Swedish cohort study of over 4.1 million individuals, identified multiple myeloma among established obesity-related cancers, reporting hazard ratios of 1.41 (1.26–1.59) for myeloid neoplasms, which includes multiple myeloma [70]. However, we found each 5 kg/m² BMI increase associated with only 2% higher MM risk, compared to 6% in previous meta-analyses [47], likely due to methodological limitations in earlier studies. For instance, the meta-analysis of Psaltopoulou et al. [48] did not include and missed several eligible studies [12, 15, 18, 20, 26, 37] and included an ineligible study [65]. Also, meta-analysis by Abar et al. [47], due to the inclusion criteria (BMI in early adulthood (aged 18–21 years)), could not include two studies [15, 23]. Surprisingly, the meta-analysis of Wallin et al. [45] and Larsson et al. [46] combined the risk estimates of plasma cell neoplasms (plasma cell myeloma and plasmacytoma) with MM, which is unacceptable according to the current WHO classification system [71]. In addition, none of the previous meta-analyses assesses the link between underweight and MM.

Obesity is associated with increased levels of inflammatory biomarkers, particularly IL-6, which is synthesized in adipocytes [6]. Previous studies have shown that IL-6 plays an important role in the proliferation and differentiation of normal and malignant plasma cells [7]. Obesity is also associated with insulin resistance and hyperinsulinemia, which can lead to an increase in IGF-I concentrations [8]. Elevated levels of IGF-1 have been shown to stimulate myeloma cell proliferation and inhibit apoptosis [9]. In addition, a high concentration of IGF-I is associated with cancer progression and poor prognosis in patients with MM [10]. In the non-linear dose-response analysis, the positive association between BMI and MM risk was stronger at a BMI of ≥ 27 kg/m². This range covers individuals with overweight and obesity. In addition, the mentioned mechanisms are amplified in these individuals compared with normal-weight subjects. This threshold of BMI ≥ 27 kg/m² is clinically significant as it encompasses individuals classified as overweight and obese, who may benefit most from weight management interventions. The stronger association at this BMI range suggests that the obesity-related inflammatory and metabolic mechanisms become more pronounced above this threshold, providing a potential target for prevention strategies.

In this meta-analysis, no significant association was found between underweight and the risk of MM and its mortality. Similar to our findings, previous studies showed no significant association between underweight and other lymphoid malignancies. For instance, two prospective cohort studies indicated no significant association between underweight and risk of non-Hodgkin lymphoma (NHL) [72, 73]. In another prospective cohort study, Fernberg et al. reported no significant association between BMI < 18.5 and the risk of Hodgkin lymphoma (HL) [74]. Such a non-significant association was also reported for NHL mortality in a prospective study [75]. The lack of a significant association in our study may be attributed to several factors. First, the number of underweight individuals in the included studies was relatively limited (only 5 studies for MM incidence and 6 studies for MM mortality), which may have resulted in insufficient statistical power to detect an association if one exists. Second, underweight status can result from various heterogeneous causes, including malnutrition, underlying diseases, or genetic factors, which could obscure any potential relationship with MM risk. Third, while underweight individuals may have lower levels of pro-inflammatory cytokines such as IL-6 and tumor necrosis factor-alpha (TNF-α) compared to overweight or obese individuals [76], which may not provide the inflammatory environment that promotes myeloma cell proliferation and survival, the etiology of MM involves multiple complex pathways beyond inflammation alone. Further well-powered prospective studies are needed to definitively assess the association between underweight and MM.

This study had some strengths. All included studies had a prospective design. Therefore, the effects of recall and selection bias, which are subjects of concern in case-control studies, can be ignored. In addition, between-study heterogeneity was low in most associations evaluated in this meta-analysis. Also, many included studies had a relatively long follow-up period and a large number of cases, which increased the statistical power of our analysis. Despite the above-mentioned strengths, our study had some limitations. First, since the studies included in the current meta-analysis were observational, we could not establish causality between BMI and MM. Second, some included studies did not control their analyses for important confounders (such as physical activity and dietary factors). Also, the risk factors for MM are unclear; we cannot rule out the possibility that the observed correlation between BMI and MM is due to unclear confounders. Third, most included studies used self-reported weight and height for BMI calculation rather than objective measurement, which may have biased our results. It has been reported that obese and overweight individuals tend to overreport height and underreport weight compared to normal-weight individuals [77]. However, when we analyzed data based on measured or self-reported anthropometry, the results were mostly similar with overlapping CIs. Fourth, some studies were not included in the dose-response meta-analysis because of insufficient data. Fifth, although no significant publication bias was detected using Egger’s test and the trim-and-fill method, the possibility of residual publication bias cannot be entirely excluded, as studies with null findings may remain unpublished or studies in non-English languages may have been missed [78, 79]. Finally, treating sex-specific results from the same cohort as independent studies may not fully account for within-cohort correlation, which could affect the precision of pooled estimates. However, the low between-study heterogeneity observed in most analyses (I²=0%) suggests that this limitation did not substantially influence our findings.

Conclusion

In conclusion, we found that high BMI, defined as obesity and overweight, was associated with an increased risk of MM and its mortality. Also, in the dose-response meta-analysis, each 5 kg/m² increase in BMI was associated with a 2% higher risk of MM and a 3% higher risk of MM mortality. However, no significant association was found between underweight and MM incidence/mortality. Based on these findings, we recommend maintaining a healthy body weight to prevent MM. Also, further studies, particularly well-designed prospective cohort studies, are needed to assess the associations of other anthropometric measures and other types of obesity with the risk of MM.

Abbreviations

MM

Multiple myeloma

BMI

Body mass index

CI

Confidence interval

CPS-II

Cancer prevention study-II

CTS

California teachers study

EPIC

European prospective investigation into cancer and nutrition

HL

Hodgkin lymphoma

HR

Hazard ration

IGF-I

Insulin-like growth factor-I

JACC

Japan collaborative cohort study

Me-Can

Metabolic syndrome, and cancer project

NHL

Non-Hodgkin lymphoma

NOS

Newcastle Ottawa scale

OR

Odds ratio

PRISMA

Preferred reporting items for systematic reviews and meta-analysis

RR

Relative risk

SCIOWE

Swedish construction industry’s organization for working environment

WHO

World Health Organization

US

United State

IL-6

Interleukin-6

TNF-α

Tumor necrosis factor-alpha

Authors’ contributions

All authors contributed to the conceptualization of the systematic review and meta-analysis. H.Z and N.E conducted the data searching. M.H and N.E conducted the data screening and quality assessment. H.Z and N.E conducted the data extraction and drafted the manuscript. M.SH and MR.J performed a meta-analysis and drafted the manuscript. All authors provided a critical review of drafts and have read and approved the final manuscript. The work reported in the paper has been performed by the authors unless specified in the text.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study is a systematic review and meta-analysis, and ethics approval and consent to participate are not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.