Relationship between monoclonal gammopathy of undetermined significance and radiation exposure in Nagasaki atomic bomb survivors

Monoclonal gammopathy of undetermined significance (MGUS) is a premalignant plasma cell disorder, which is defined by a serum monoclonal protein (M-protein) concentration of 3 g/dL or less, 10% or fewer plasma cells in the bone marrow, and the absence of anemia, osteolytic lesions, hypercalcemia, and renal dysfunction.¹  Although the majority of patients with MGUS remain stable for prolonged periods, malignant transformation to multiple myeloma occurs at a constant rate of 1% per year.²  Given that myeloma is an incurable hematologic malignancy, it is important to elucidate etiology and predisposing factors of MGUS.

Etiologic factors for MGUS have not been investigated fully.^3-5  There are currently no consistent risk predictors beyond age, sex, and race for developing MGUS. Although radiation exposure is well known to initiate leukemogenesis, there have been conflicting reports about the association between radiation exposure and plasma cell disorders.^6-11  An Italian case reference study reported an increased risk of MGUS among people exposed to occupational radiation exposure.¹²  However, a small survey for atomic bomb survivors showed no association between radiation dose and the relative risk of MGUS.¹³  Sample sizes of these previous studies were too small to obtain reliable results for association between radiation exposure and incidence of the disease.

We have recently reported the age-specific MGUS prevalence in the Japanese population, indicating 2.4% in those older than 50 years.¹⁴  The report used an M-protein screening data from approximately 52 000 atomic bomb survivors but did not report the relationship between radiation exposure and MGUS risk. The large number of study participants from the radiation-exposed population could provide a great opportunity to investigate the relationship between radiation exposure and the risk of MGUS. Our preliminary analysis observed that MGUS risk was higher in those exposed to higher radiation at a young age.¹⁵  However, the preliminary observation lacked detailed analyses for the relationship and did not include clinical characteristics. In the present study, we performed comprehensive analyses for the screening data by considering distance from the hypocenter of the nuclear explosion, radiation dose, age at exposure, age at diagnosis, and M-protein level to elucidate whether radiation exposure is related with the development of MGUS and the progression.

Screening for M-protein was initiated in October 1988 for atomic bomb survivors at the Health Management Center of Nagasaki Atomic Bomb Casualty Council, where a comprehensive medical check-up has been offered twice a year since 1968; several cancer screenings have been offered once a year since 1988. All examinations are free of charge and supported by the Nagasaki City government based on the Law Concerning the Relief for Atomic Bomb Survivors. Data of all medical check-ups and cancer screenings have been stored online in the computer database at the Data Center in the Atomic Bomb Disease Institute at Nagasaki University Graduate School of Biomedical Sciences since 1977. The ongoing database keeps data from approximately 120 000 atomic bomb survivors who have the Atomic Bomb Victim's Handbook, including fundamental information, age at exposure, city at the time of the bombings, exposure categories, exposure distance from the hypocenter in kilometers, date of the certificate handbook acquisition, date of examination, date of death, and date of moving to or away from Nagasaki City, and all laboratory results.¹⁶  Data of participants who underwent the M-protein screening were extracted as anonymous data from the computer database in the Data Center of Atomic Bomb Disease Institute. Use of the database for this study was approved by the Atomic Bomb Disease Institute in June 2004 (No. 224).

Screening procedures were described in detail previously.¹⁴  Briefly, routine laboratory tests including the first-step M-protein screening were offered every year for atomic bomb survivors who visited the Health Management Center. Results of the first-step M-protein screening were evaluated on the sheet in the double-checking system by hematologists of Nagasaki University Medical Hospital regardless of exposure condition. Survivors with the presence of possible M-protein or low gammaglobulinemia were informed by mail or telephone to take the second-step screening. The second-step screening procedure consisted of physical examination by hematologists, immunoelectrophoresis of serum and urine, a qualitative test for Bence-Jones (BJ) protein, and a quantitative determination of serum concentration of immunoglobulins (Igs) using nephelometry. Subjects with a high level of M-protein or with other abnormal laboratory data were referred to the tertiary hospitals to undergo further examination with bone marrow aspiration, bone surveys, and other investigations. Skilled hematologists made a final diagnosis comprehensively based on screening data, routine laboratory data, physical examinations, and feedback letters from the reference hospitals. The diagnostic criteria used for MGUS were based on an M-protein level lower than 3.0 g/dL in serum Igs, no symptom of multiple myeloma or Waldenström macroglobulinemia (WM), no anemia, no hypercalcemia, no osteolytic lesion, and less than 10% marrow plasma cells if done.¹  Patients with high M-protein levels of more than 3.0 g/dL on the first-time detection day but showing the “reconfirmed” M-protein levels of less than 3.0 g/dL were also treated as MGUS.

Radiation exposure was assessed by exposure distance and exposure dose. In the database, exposure categories were divided into 4 categories: “directly exposed” indicates those who were exposed to atomic bomb radiation within 10 km from the hypocenter at the time of the bombing; “early entrants” indicates those who entered the city within approximately 2 km from the hypocenter within 2 weeks of the explosion; “relief” indicates those who were engaged in disposal of the dead or relief works for atomic bomb victims; and “exposed in utero” indicates children who were exposed prenatally at the time of the bombing. Although information of exposure distance was available for “directly exposed” and “exposed in utero,” we used only “directly exposed” people for the analysis to investigate the relationship between radiation exposure and MGUS risk. Information on whole-body radiation dose estimates by the Atomic Bomb Survivors 1993 Dose (ABS93D) was available for a limited number of Nagasaki atomic bomb survivors.¹⁷  The ABS93D was calculated according to 3 parameters: free-in-air kerma, shielded kerma, and organ kerma, which is the same as Dosimetry System (DS) 86,¹⁸  which was used for the Life Span Study (LSS) cohort of the Radiation Effects Research Foundation (RERF).¹¹  Because a strong correlation between DS86 and ABS93D was fully documented,¹⁹  we used ABS93D as a substitute for the DS series to estimate radiation-dose response.

The target population for the M-protein screening was 74 411 atomic bomb survivors exposed in Nagasaki City, consisting of 71 675 people who were alive at the time of the start of the M-protein screening in October 1988 and 2736 people who were included in the database after 1988 to 2004 because some newly obtained the Atomic Bomb Victim's Handbook and others moved to Nagasaki City from elsewhere. Those exposed in Hiroshima City who moved to Nagasaki City were excluded. Among 74 411 people, ABS93D dose information was available for 6837 (9.2%). Table 1 presents the breakdown of participants and nonparticipants by demographic characteristics. The participant rates were around 70% in all categorized groups except in those aged 30 years or older at exposure (36%). Finally, a total of 52 525 Nagasaki atomic bomb survivors underwent the M-protein screening between 1988 and 2004 (overall participation rate, 70.6%) and were used for analyses to examine the relationship between MGUS risk and exposure distance from the hypocenter. Among those with information on ABS93D dose, 4758 (participation rate, 69.6%) underwent the screening and were used for the dose-response analyses.

Participants who were once diagnosed as having MGUS also annually underwent the M-protein screening in the same way as described in the screening procedure. They were followed to check the change in M-protein levels on an individual M-protein chart, which was reviewed by skilled hematologists in the Health Management Center. Participants with a high level of M-protein or with other abnormal laboratory data were referred to the tertiary hospitals to undergo further examination. Diagnoses of multiple myeloma or other related diseases were obtained from the tertiary hospitals.

Statistical analyses for prevalence were performed using all the screening data accumulated during the period from October 1, 1988, to March 31, 2004. Patients who were diagnosed with multiple myeloma or WM at the first-time screening were excluded from the analyses. Patients with MGUS were also analyzed for the risk of malignant progression during the period from the date of diagnosis to July 31, 2008. All statistical analyses were performed using SAS 8.2 software (SAS Japan Institute, Tokyo, Japan). All tests were 2-tailed, and the level of statistical significance (P) was .05.

Age at exposure was treated as a continuous data or stratified into 4 categories (0-9 years, 10-19 years, 20-29 years, and 30 years or older). Exposure distance from the hypocenter in kilometers was treated as a continuous data or stratified into 3 categories (within 1.5 km, 1.5-3.0 km, and 3.0 up to 10.0 km). The cutoff values for exposure distance were chosen based on previous reports.^20,21  Among those in the exposure category of “directly exposed,” participants with no information on distance position were treated as those exposed at an unknown distance. The ABS93D dose estimate in grays was treated as continuous data or stratified into 3 categories (lower than 0.01 Gy, 0.01-0.1 Gy, and 0.1 Gy or higher). Age at diagnosis of MGUS was stratified into 5 categories (< 50 years, 50-60 years, 60-70 years, 70-80 years, and ≥ 80 years). Basic demographic analyses were assessed using the χ² test or trend test for categoric variables and the nonparametric test for continuous variables, if necessary. Simple prevalence (percentage) of MGUS and the 95% confidence intervals (CIs) were calculated using the exact binomial method in each category. Exposure-response analyses were performed for 2 datasets, one for people with assured exposure distance from the hypocenter to examine the relationship between MGUS risk and the exposure distance, and another for people with assured ABS93D dose to examine the relationship between MGUS risk and exposure dose. To evaluate the relationship between MGUS risk and exposure distance or exposure dose, we calculated prevalence ratios (PRs) and the 95% CI with the log-binomial regression model using PROC GENMOD in SAS.^22,23  Univariate and multivariate analyses were performed, including relevant factors and/or interaction terms to test effect modification. To obtain the best-fit model for dose-response effect, we ran additional analyses, including sex, continuous age at exposure per year, continuous radiation dose (linear or quadratic term), and interaction terms between covariates. The most appropriate model was selected on the basis of the Akaike Information Criterion (AIC).²⁴  The cumulative probability of developing multiple myeloma or other lymphoid malignancies among patients with MGUS was calculated using the Kaplan-Meier method and compared using the log-rank test. Patients who died or were lost to follow-up were censored in the analysis.

Of 52 525 participants, 1103 were confirmed as having monoclonal immunoglobulin, of which 1082 were diagnosed with MGUS, 19 with multiple myeloma, and 2 with WM. The 21 patients with multiple myeloma or WM were excluded from analyses. Therefore, a total of 52 504 participants were used for analyses by exposure distance. Of the 21 patients excluded, 3 had ABS93D dose information. Therefore, a total of 4755 participants with ABS93D dose were used for dose-response analyses.

Table 2 shows the clinical characteristics of 1082 patients with MGUS. The median age at diagnosis was 68.5 years (range, 45.0-100.9 years). Age at diagnosis was significantly older in women (median, 68.3 years) than in men (median, 66.3 years; P = .003). The distribution of age at diagnosis by exposure categories is presented in Table 3. Although patients exposed at a younger age tended to be younger at diagnosis (Figure 1C), there was no difference in age at diagnosis across exposure distance groups (P = .65); however, there was some tendency for age at diagnosis to be younger in those exposed to the higher dose (> 0.1 Gy) than those exposed to the lower dose in each exposure age group, though the differences were not statistically significant (P = .46 among 3 dose categories and P = .23 between the dose group of 0-0.01 and > 0.1 Gy; Figure 1A,B). Median serum M-protein level at diagnosis was 1.5 g/dL (range, 0.1-3.4 g/dL). The distribution of serum M-protein level by demographic characteristics is summarized in Table 4 and Table S1 (available on the Blood website; see the Supplemental Materials link at the top of the online article). MGUS with M-protein levels greater than 1.5 g/dL were highly frequent in those exposed at 20 years or older. However, the level was not different among age at diagnosis, exposure distance, or exposure dose.

MGUS prevalence in 52 504 participants by sex and exposure status is shown in Table 5. The overall prevalence of MGUS in participants was 2.1% (95% CI, 1.9-2.2), 2.8% (95% CI, 2.6-3.0) in men, and 1.6% (95% CI, 1.5-1.7) in women. MGUS prevalence was 2.7% (95% CI, 2.1-3.4) in those directly exposed within 1.5 km from the hypocenter, 1.9% (95% CI, 1.7-2.2) at 1.5 to 3.0 km, 2.0% (95% CI, 1.8-2.1) at more than 3.0 km, and 2.3% (95% CI, 2.0-2.6) in other exposure categories. Table 6 summarizes the results of univariate and multivariate regression analyses. The unadjusted PR was significantly higher in men, in those at an older age at exposure, and in those exposed at within 1.5 km compared with those exposed at more than 3.0 km. A multivariate analysis including interaction terms among all variables showed a significant interaction (P < .03) between age at exposure and the exposure distance, but no significant interaction between sex and age at exposure (P < .7) or exposure distance (P < .9), suggesting that the effect of exposure distance on MGUS prevalence might be different by age at exposure. Therefore, we analyzed data by dividing into 2 age categories: those exposed at younger than 20 years and those exposed at age 20 years or older. Because age function is a strong risk factor for MGUS, we included age at exposure as a continuous variable in both stratified multivariate analyses. In the multivariate analysis for those exposed when younger than 20 years, the adjusted PR of MGUS showed a 40% increase per every 5-year increase of age at exposure (adjusted PR, 1.4; 95% CI, 1.3-1.5), and the probability of MGUS among participants who were exposed within 1.5 km was overall 40% higher than among those exposed further than 1.5 km (adjusted PR, 1.4; 95% CI, 1.1-1.9). The adjusted PR of MGUS showed no difference among exposure distance categories in those of age at exposure older than 20 years.

Table 7 presents the breakdown of MGUS prevalence in people with information on ABS93D dose. Among dose categories, the prevalence was 2.5% (95% CI, 1.7-3.5) in those exposed to 0.1 Gy or more, 2.0% (95% CI, 1.4-2.8) in those exposed to 0.01 to 0.1 Gy, and 1.6% (95% CI, 1.1-2.3) in those exposed to 0.01 Gy or less. Before applying dose as continuous data, doses are truncated to correspond to the 4 Gy level according to previous RERF studies.^11,25 Table 8 summarizes results of univariate and multivariate regression analyses for PRs. For those exposed when younger than 20 years, univariate analyses showed significantly higher PR in those exposed to 0.1 Gy or more compared with those exposed to a lower-dose category. However, no significant dose effect was observed when dose was treated as a continuous variable. After adjusting sex and age at exposure, the PR of MGUS in those exposed to 0.1 Gy or more was estimated at 1.66, suggesting that radiation exposure over 0.1 Gy had a 1.66 times higher risk of MGUS compared with the dose of less than 0.1 Gy. However, the linear-dose model failed to find a clear dose-response effect even after controlling sex and age at exposure (multivariate analysis 1 in Table 8). We performed additional models, including dose as a treated quadratic transformation. The AIC value in each multivariate analysis was 587.7646 for a model using a linear term (the parameter estimate [beta] for dose, 0.2179; standard error [SE], 0.1651; P = .2), 588.0652 for a simple quadratic term (dose squared; beta, 0.0569; SE, 0.0469; P = .2), and 589.7468 for a quadratic term (beta, 0.2794; SE, 0.4867; P = .5). For those exposed when older than 20 years, both univariate and multivariate analyses showed no effect of radiation dose on MGUS prevalence even after controlling other covariates. Figure 2A shows the PR of MGUS by exposure dose squared, adjusting for sex and age at exposure. PR at 1 Gy was 1.06 (95% CI, 0.97-1.16; P = .2) among those aged younger than 20 years at exposure. Figure 2B shows the PR of MGUS by age at exposure adjusting for sex and exposure dose squared. Advanced age was significantly associated with an increased prevalence of MGUS among those aged younger than 20 years at exposure (PR, 2.24 for 10-year increase; 95% CI, 1.39-3.62; P = .001) and those older than 20 years (PR, 1.77 for 10-year increase; 95% CI, 1.03-3.03; P = .04).

Patients with MGUS were followed for a total of 8822.5 person-years (median, 7.4 years; range, 0-19.6 years). There were 365 (33.7%) patients who were followed until death. During this period of observation, 44 (4.1%) patients experienced the progression to multiple myeloma (41 patients) and WM (3 patients). All cases of myeloma were developed from IgG or IgA MGUS. Among 3 cases of WM, 2 were developed from IgM MGUS, and one was developed from IgG MGUS.²⁶  The median latency period between the diagnosis of MGUS and the development of multiple myeloma or WM was 5.3 years (range, 0.1-15.9 years). The overall cumulative probability of the progression was 6.9% (95% CI, 4.9-9.6) at 10 years and 8.0% (95% CI, 5.4-11.9) at the latest follow-up (Figure 3A). Among the 44 patients, 36 had information on exposure distance, and only 2 had information on exposure dose. Therefore, risk analyses were performed only by exposure distance. The frequency of malignant progression by factors is summarized in Table 9. The cumulative probability of the progression was greater in those exposed within a 1.5-km distance than those exposed at 1.5 to 3.0 km and at 3.0 km or more, but the difference was not statistically significant (13.9% vs 6.7% vs 7.7%; log-rank test P = .34; Figure 3B). The probability was significantly higher for those exposed at 20 years of age and older than those exposed at younger than 20 years (18.1% vs 5.4%; P = .04; Figure 3C). Among those exposed at 20 years old or older, there was no difference in the progression between those exposed within 3 km and greater than 3 km from the hypocenter (P = .90), but among those aged younger than 20 years at exposure, the probability had a tendency to be high in those exposed at within 3 km than those exposed more distantly (7.4% vs 4.2%; P = .17; Figure 3D). Among those aged 20 years or older at exposure, those diagnosed younger than the median age of 68.5 years significantly progressed to myeloma more than those older than 68.5 years (35.4% vs 7.6%; P = .02; Figure 3E). The cumulative probability was significantly higher in those with higher M-protein levels at diagnosis (≥ 1.5 g/dL) than those with lower levels (< 1.5 g/dL; 12.5% vs 2.0%; P < .001; Figure 3F). The older age at exposure showed the greater risk of progression among those with higher M-protein levels at diagnosis (P = .06), but there was not a different risk in age categories among those with the lower M-protein levels (P = .80; Figure 3G). There was no risk difference between the exposure distance categories among those with higher M-protein levels at diagnosis (P = .60), but there was a tendency for greater risk in those exposed within 3 km of the hypocenter among those with lower M-protein levels (P < .001; Figure 3H).

The present study is the first comprehensive evaluation of the effects of radiation exposure on MGUS prevalence using a large number of atomic bomb survivors. We observed that, among those exposed when younger than 20 years, the probability of MGUS was 1.4 times greater in those exposed near the hypocenter than those exposed far from the hypocenter, and 1.7 times greater in those exposed to radiation doses of 0.1 Gy or more than those exposed to smaller radiation doses. We also observed that the strongest factor on the progression of MGUS was the high level of M-protein at diagnosis rather than the effect of the higher radiation exposure.

Only a few epidemiologic studies reported an effect of radiation exposure on MGUS. Pasqualetti et al observed that occupational exposure to radiation was significantly associated with an increasing risk of MGUS.¹²  However, the result was based on only 13 patients, and no dose-response analysis was performed. Neriishi et al reported no association between radiation dose (DS86) and the incidence of 112 (1.7%) cases of MGUS among 6737 atomic bomb survivors who were members of the Adult Health Study (AHS) of RERF.¹³  The study found that the MGUS risk was not different between those exposed to more than 0.01 Gy and those exposed to less than 0.01 Gy (relative risk [RR] = 1.35; 95% CI, 0.9-2.0). There were several differences between the AHS study and the present study in terms of analytic method and observed results. The overall prevalence was lower in the AHS study than our result (1.7% vs 2.1%) despite the same study periods. The cut-off value to compare MGUS risk by dichotomized dose category was also different, as the present study uses 0.1 Gy, but the AHS study used 0.01 Gy. This difference might affect the different interpretation of the results. In addition, the AHS study did not observe the significant interaction between age at exposure and dose, nor demonstrated dose-response analysis by age at exposure. Nevertheless, they realized marginally significant increases in MGUS risk in those younger than 80 years at onset, which might support our result that a significantly higher prevalence risk of MGUS was observed only in participants who were exposed at a younger age. Even though there were some differences between the AHS study and our present study, the estimated MGUS risk was similar; the RR was 1.603 (P = .05) in those younger than 80 years at diagnosis in the AHS study, and the PR was 1.66 (P = .05) in those exposed when younger than 20 years in our study. This suggests that it is consistent that there exists a significant weak association between radiation exposure and MGUS risk among those exposed when young.

Although we found that only younger age at exposure had a significant association between the higher-dose radiation exposure and the higher MGUS risk, the result does not necessarily deny the association in those of older exposure age. As shown in Table 1, the participation rate was lower in older ages, which suggests the data are less representative of the actual MGUS prevalence among the older target population. As is well known, older atomic bomb survivors, especially those exposed at the higher radiation dose, had higher mortality due to both cancers and noncancer diseases.^21,27  Therefore, results among those exposed at 20 years of age or older in our study might be strongly affected by detection loss.

For the association between radiation and myeloma, a number of epidemiologic studies analyzed people exposed to environmental, occupational, and medical radiation.^5,28,29  A series of reports from Hanford nuclear workers in the US and Sellafield workers of British Nuclear Fuels indicated a significant dose-response trend between death from myeloma and cumulative external radiation dose.^30-35  A recent international report of the 15-country collaborative study of nuclear workers also found a borderline significant association with radiation dose and 87 deaths from myeloma (RR = 1.61 at 100 mSv).³⁶  The age effect in most nuclear worker studies reported that a significant dose response was observed in those of older ages at exposure, which differs from findings in the experiences of atomic bomb survivors, including our study, that significant dose responses were observed to be more likely in those who were younger at exposure. For this discrepancy, Wing et al discussed that selection bias and basic differences in the characteristics of the study populations may be considered.⁷  Another difference might be due to the differences in the type of exposure to radiation; nuclear workers received chronic exposures to cumulative lower doses over their lifetimes, in contrast with atomic bomb survivors, who received acute exposure to high doses of radiation.

Unlike nuclear workers, there is no epidemiologic evidence supporting an increased risk of myeloma among atmospheric nuclear test participants.^8,9,37-40  All of these studies had less power to evaluate dose-response association because the observed number of myeloma cases was too small (less than 8). A mixed association has been observed between risk of myeloma and diagnostic or therapeutic radiation.^41-43  In a large international study of radiation treatment for cervical cancer, there was no difference in risk of myeloma between those who were received less than 2 Gy and 2 Gy or greater; however, increased risks were observed among patients followed long term and those irradiated at relatively younger ages.⁴⁴  The observation supports our result that the higher MGUS risk was observed in those exposed to the higher radiation at a younger age.

Among atomic bomb survivors, the relationship between exposed radiation dose and myeloma has been also inconsistent. Ichimaru et al analyzed 29 patients with myeloma accumulated between 1950 and 1976, and found a statistically significant increase in the incidence among the higher-dose group (more than 0.5 Gy) since 1965, suggesting a prolonged latency period for radiation-induced myeloma.¹⁰  The study also indicated a different dose effect by age of exposure; the positive effect was seen only in those aged 20 to 59 years at exposure, which was very similar to our present study. Shimizu et al also reported a statistically significant excess risk for myeloma from 1950 to 1985.⁴⁵  However, the latest report did not observe a significant dose response (P = .12) when analyses were limited to first-primary myeloma cases, though a statistically significant increase was observed when excluded cases were included in the analysis (P = .02).¹¹  In the latest report, only 59 of 94 patients were used for the analysis because many patients were excluded for a variety of reasons. The report explained the discrepancy within the same cohort might be affected by differences in the inclusion criteria of case and dosimetry system.

The majority of patients with MGUS will never develop MM. So far, the size of serum M-protein, the IgA isotype, an abnormal serum free light-chain ratio, detectable BJ protein excretion, and more than 5% of plasma cells in bone marrow have been identified as predictors of MM progression.^46-48  Nevertheless, precise predictors to define patients with high-risk MGUS should be identified. In the present study, we confirm that the strongest factor on the progression of MGUS is the high level of M-protein at diagnosis rather than the effect of radiation exposure. Exposure age and age at diagnosis showed complicated effects on the prognosis. Those exposed at 20 years or older progressed more than those exposed at younger ages (Figure 3C), but those diagnosed younger than 68.5 years were more likely to progress to myeloma in both exposure age categories (Figure 3E). These results might be affected by the competing cause of death; the older patients would die before the progression of MGUS, which could introduce the underestimate of the progression risk among older patients. Although the present study did not find confident evidence that radiation exposure was related to the malignant progression of MGUS, there was a tendency for a greater risk of progression among patients exposed proximally. Neriishi et al also reported that the multiple myeloma mortality rate was higher among the exposed group (> 0.01 Gy, 10 patients) than the nonexposed group (0-0.01 Gy, 4 patients), though the difference was not significant.¹³  Both studies suggested a potential adverse effect of radiation exposure on the progression from MGUS to multiple myeloma.

The present study has several limitations. Dose analyses were performed for a limited number of subjects. A healthy screenee bias⁴⁹  might affect the results, especially in older age group. Indeed, the participation rate decreased by age (Table 1). Overdiagnosis bias surely exists because of the long-term prognosis of MGUS in nature. Potential factors included in analyses were also insufficient. These limitations would have introduced over- or underestimates of the association. Further researches including other potential factors as covariate together are needed to confirm the effect of radiation on MGUS.

The mechanism of how radiation exposure affects the increasing risk of MGUS is still unknown. As is well known, radiation exposure induces chromosomal and genomic instabilities by direct and indirect ways.⁵⁰  Meanwhile, a variety of chromosome abnormalities have been reported even though MGUS is a benign hematologic disorder.^51,52  These facts might explain that MGUS risk increases when exposed to the higher level of radiation dose through radiation-induced chromosomal and genomic instabilities. Beyond the effect of radiation on MGUS risk, recent epidemiologic studies provided clear evidence of a significant racial disparity in MGUS prevalence^3,4,14  and familial aggregation for multiple myeloma/MGUS,⁵³  both of which suggest a role for genetic susceptibility as MGUS etiology. More recently, Brown et al reported a possible role for immune-related and inflammatory conditions in the causation of MGUS.⁵⁴  This report may also suggest another perspective on radiation-induced MGUS because recent molecular studies have revealed that radiation-induced inflammatory reaction and radiation-induced genomic instability may be interrelated with a predisposition to radiation carcinogenesis.^50,55 

We previously reported that, even allowing for atomic bomb survivors, our Japanese population had a lower prevalence of MGUS compared with the white population.¹⁴  Although the conclusion is solid evidence, the present findings suggest that the prevalence data of atomic bomb survivors may not be generalizable to other Japanese populations, but rather suggest that MGUS prevalence in a general Japanese population might be lower than our population because the present study showed that those who were exposed to lower radiation had a significantly lower prevalence. Further population-based epidemiologic studies using the general population are needed to estimate a more reliable MGUS prevalence in Japanese and other Asians.

In conclusion, the present study suggests that atomic bomb survivors exposed to high levels of radiation at young ages are at high risk of the evolution of MGUS, even many years after radiation exposure. During the screening period from 1988 to 2004, the population of atomic bomb survivors became older, with the youngest atomic bomb survivors around 60 years of age. Unlike leukemia, the risk of solid cancers following exposure to ionizing radiation becomes manifest after a relatively long latency period,²⁷  after which the excess risk persists for decades. MGUS and myeloma are also diseases with long latencies. Further investigations of MGUS and myeloma are needed for this large and long-followed population, especially for people exposed at younger ages.

We are grateful to many hematologists of the Nagasaki University Medical Hospital and staff at the Health Management Center of Nagasaki Atomic Bomb Casualty Council involved in the daily M-protein screening procedures.

This study was supported by a research grant from the Ministry of Education, Culture, Sports, Science and Technology of Japan for the Radiation Medicine Program of Nagasaki University 21st Century Research Centers of Excellence (No. 17301, E-17, Project Grant to M. Tomonaga). The sponsors had no role in the analysis, the interpretation of the results, the preparation of the manuscript, or the decision to submit the manuscript for publication. All medical examinations for A-bomb survivors are charge-free based on medical aid support from the Nagasaki City government.

Contribution: M.I. was involved in the screening procedure, analyzed data, and wrote the manuscript; M. Tagawa established and managed the screening procedure; T.M. managed the screening procedure; K.Y. administrated and extracted data from the Data Center in the Atomic Bomb Disease Institute; Y.M., T.F., T.H., Y.I., D.I., and J.T. were responsible for the first screening procedure; S.K., S.M., and K.T. were responsible for the final diagnosis of the screening procedures; M. Tomonaga established the screening procedure and managed the database of the Atomic bomb survivors; and all authors revised the article critically and approved the final version.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Masako Iwanaga, Department of Hematology and Molecular Medicine, Atomic Bomb Disease Institute, Nagasaki University Graduate School of Biomedical Sciences, 1-12-4 Sakamoto, Nagasaki, 852-8523, Japan; e-mail: masakoiw@nagasaki-u.ac.jp.