Pancreatic Irradiation Dose as a Risk Factor for Elevation in Hemoglobin A1c Level and Diabetes Mellitus in Patients with Indolent Gastroduodenal Lymphoma
Article information
Abstract
Purpose
This study investigated the association between pancreatic irradiation dose and hemoglobin A1c (HbA1c) elevation in patients with indolent gastroduodenal lymphomas treated with radiotherapy (RT).
Materials and Methods
We retrospectively reviewed 103 patients treated at Samsung Medical Center between 2010 and 2023, with or without RT. Patients were stratified by mean pancreas dose (< 2,300 cGy vs. ≥ 2,300 cGy). The primary outcome was the 2-year HbA1c elevation, with additional time-to-event analyses for HbA1c elevation ≥ 0.5% and diabetes mellitus (DM) development.
Results
RT was administered to 62 (60.2%) patients, while 41 (39.8%) did not receive RT. There was no significant difference in baseline characteristics except for cancer type. Patients with a mean pancreas dose ≥ 2,300 cGy had a significantly greater 2-year HbA1c increase than those receiving < 2,300 cGy (p=0.003) or no RT (p < 0.001). No significant difference was found between patients receiving < 2,300 cGy and those without RT (p=0.120). In multivariate time-to-event analysis, a mean pancreas dose ≥ 2,300 cGy was the sole significant risk factor for an HbA1c increase ≥ 0.5% (p < 0.001), while mean pancreas dose < 2,300 cGy did not show a significant effect (p=0.851). Furthermore, a mean pancreas dose ≥ 2,300 cGy (p=0.017) and baseline prediabetes (HbA1c ≥ 5.7%) (p=0.023) were independent predictors of DM development.
Conclusion
A mean pancreas dose ≥ 2,300 cGy was associated with HbA1c elevation, whereas doses < 2,300 cGy had minimal effect. Given the established link between higher HbA1c level and increased risk of DM and cardiovascular events, pancreatic dose reduction should be considered in RT planning for patients with an expected good prognosis.
Introduction
Stage IE gastroduodenal lymphomas, specifically mucosa-associated lymphoid tissue lymphomas (MALTomas) and follicular lymphomas localized to the stomach and duodenum, are typically managed with definitive radiotherapy (RT) [1-4]. Patients with MALTomas receive definitive RT when Helicobacter pylori eradication therapy is ineffective or when they are H. pylori–negative. The standard treatment for stage IE duodenal follicular lymphoma is definitive RT. Given the excellent treatment outcomes and high survival rates of patients with these lymphomas, the treatment focus has shifted from efficacy to minimizing adverse effects [1,5-10].
Traditionally, the liver and kidneys are prioritized as organs at risk (OARs) in RT planning, while the pancreas is rarely considered [11,12]. Abdominal irradiation is reportedly associated with an increased diabetes mellitus (DM) risk in patients with childhood cancer, Hodgkin lymphoma, and testicular cancer, suggesting a relationship between the pancreatic irradiation dose and DM risk in young patients [13-16]. However, limited data in adults have kept the pancreas from being considered an OAR [11,12].
Hemoglobin A1c (HbA1c) is widely used for DM diagnosis and monitoring as it reflects the average blood glucose level over the previous 2-3 months. Unlike other blood glucose tests, which are influenced by factors such as food intake, HbA1c is a stable and reliable measure of blood glucose levels [17]. Elevated HbA1c levels are strongly associated with an increased DM risk, suggesting a continuous relationship between HbA1c levels and subsequent DM development [17,18]. Moreover, in non-diabetic individuals, higher HbA1c levels have been linked to an elevated risk of cardiovascular events [19,20].
We retrospectively analyzed HbA1c levels 2 years after treatment to determine whether differences emerge in patients receiving pancreatic irradiation, with a focus on assessing whether the pancreatic irradiation dose is related to subsequent HbA1c elevation. By analyzing potential dose-response effects on HbA1c levels at a single time point, we aimed to provide reliable insights into the elevation of HbA1c and DM risk associated with pancreatic irradiation in adult patients treated for indolent gastroduodenal lymphomas.
Materials and Methods
1. Patients
We retrospectively reviewed 153 patients with stage IE gastroduodenal MALToma or follicular lymphoma who were treated with definitive RT and/or only with H. pylori eradication at Samsung Medical Center between January 2010 and March 2023 and had baseline HbA1c test results. After excluding 50 patients who did not undergo HbA1c testing 2 years after treatment and one patient without a recorded body mass index (BMI), 103 patients were included in the analysis. The HbA1c tests in the included patients were performed either during annual health check-ups or as part of routine follow-up after RT.
2. Radiotherapy
Patients with stage IE gastric MALToma who were initially H. pylori positive received eradication therapy. Definitive RT was administered if eradication therapy was unsuccessful, as upfront therapy in some patients with H. pylori–negative MALToma, and as direct treatment in patients with duodenal MALToma or follicular lymphoma. Before April 2020, the MALToma treatment regimen was 30 Gy in 15 fractions; after April 2020, it was revised to 24 Gy in 12 fractions. Patients with duodenal follicular lymphoma received 24 Gy in 12 fractions. RT was administered in 2-Gy fractions once daily for 5 days per week. An involved site radiotherapy approach was followed; the clinical target volume encompassed the entire stomach or duodenum. All patients underwent four-dimensional respiratory phase–binned computed tomography, with treatment planning performed either in terms of a specific respiratory phase (using gating) or for all respiratory phases. Three-dimensional conformal or intensity-modulated RT was administered. The pancreatic irradiation dose was evaluated based on the mean dose delivered to the entire pancreas (mean pancreas dose).
3. HbA1c
Baseline HbA1c level was defined as the level recorded within 6 months before or after treatment initiation. The 2-year HbA1c level was defined as the level recorded within 6 months before and after the 2-year point following treatment initiation. The difference between these two levels was termed the “2-year elevation in HbA1c levels,” and we primarily evaluated differences in this parameter according to the pancreatic irradiation dose.
Additionally, we conducted a time-to-event analysis to assess the probability of an elevation in HbA1c level of ≥ 0.5% or DM (HbA1c level ≥ 6.5%) throughout the follow-up period. A change of 0.5% is generally considered significant [21]. The time to an elevation in HbA1c level of ≥ 0.5% was defined as the interval from baseline HbA1c testing to the first instance of an elevation in HbA1c level of ≥ 0.5%. The time to DM was defined as the interval from baseline HbA1c testing to the first instance of an HbA1c level of ≥ 6.5%. A total of 445 HbA1c test results since baseline were analyzed. The intervals between HbA1c tests were 1 year for 393 tests, 2 years for 49 tests, 3 years for 2 tests, and 4 years for 1 test. Three- to four-year intervals were observed during the coronavirus disease 2019 pandemic. The subsequent HbA1c values of patients who started using medications that could affect glucose levels during the follow-up period were excluded from further analysis.
4. Statistical analysis
Fisher’s exact test was used for categorical variables, while independent t-tests were applied to compare continuous variables. To assess the association between mean pancreas dose and 2-year elevation in HbA1c level, linear regression analysis was performed.
Due to the non-normal distribution of HbA1c elevation, the Mann-Whitney U test was applied when comparing 2-year elevation in HbA1c level across various variables. The 2,300 cGy cutoff was determined based on the lowest p-value. Additionally, a subgroup analysis was performed by excluding patients with prediabetes at baseline (HbA1c 5.7%-6.4%), and the association between mean pancreas dose and 2-year elevation in HbA1c level remained statistically significant with the 2,300 cGy cutoff still showing the lowest p-value. All other potential confounding factors (age, sex, obesity (BMI ≥ 25 kg/m2), hypertension, dyslipidemia, and baseline HbA1c level) had p-values > 0.10, so a multivariate model was not considered.
For time-to-event analyses, univariate and multivariate Cox regression models were used, and variables with p < 0.10 in the univariate analysis were included in the multivariate model for adjustment. Kaplan-Meier curves with log-rank tests were generated.
Statistical significance was set at p < 0.05 (two-sided). All analyses were conducted using SPSS ver. 27 (IBM Corp.).
Results
1. Participant characteristics
Of the 103 patients, 62 (60.2%) received RT and 41 (39.8%) did not. In the RT group, 80.6%, 1.6%, and 17.7% of patients had gastric MALToma, duodenal MALToma, and duodenal follicular lymphoma, respectively, while all no-RT patients had gastric MALToma (p < 0.001). There was no significant difference in other baseline characteristics (Table 1). Among patients who received RT, 11 (17.7%) received 30 Gy and 51 (82.3%) received 24 Gy. The 24 Gy group had a significantly lower mean pancreas dose than the 30 Gy group (median [interquartile range (IQR)], 1,863 cGy [1,594 to 2,081 cGy] vs. 2,619 cGy [2,395 to 2,900 cGy], p < 0.001) (S1 Table).
Baseline characteristics of patients with stage IE indolent gastroduodenal lymphoma (n=103)
2. 2-Year HbA1c elevation
Fig. 1 shows the 2-year HbA1c elevation by mean pancreas dose. Higher doses correlated with greater HbA1c increases (linear regression analysis, β=0.391, R2=0.15, p=0.002). The optimal cut-off value for mean pancreas dose was determined to be 2,300 cGy as it yielded the smallest p-value (S2 Table). A subgroup analysis excluding patients with prediabetes (HbA1c 5.7-6.4) at baseline confirmed a significant association between mean pancreas dose and 2-year elevation in HbA1c level in linear regression analysis (p=0.023). Additionally, the 2,300 cGy cutoff remained the optimal threshold, yielding the lowest p-value (p=0.013).
Distribution of the 2-year elevation in hemoglobin A1c (HbA1c) levels according to mean pancreas dose in patients who received radiotherapy, with a linear regression line (p=0.002).
Patients in the ≥ 2,300 cGy group exhibited a significantly greater increase in HbA1c levels than the < 2,300 cGy group (median [IQR], 0.25 [0.20 to 0.45] vs. 0.10 [0.00 to 0.20]; p=0.003). Furthermore, patients in the ≥ 2,300 cGy group exhibited a significantly greater increase in HbA1c levels compared to the no RT group (median [IQR], 0.25 [0.2 to 0.45] vs. 0.00 [−0.10 to 0.20]; p < 0.001). However, no significant difference was observed between the < 2,300 cGy group and the no RT group (median [IQR], 0.10 [0.00 to 0.20] vs. 0.00 [−0.10 to 0.20]; p=0.120) (Fig. 2).
Distribution of the 2-year elevation in hemoglobin A1c (HbA1c) level in patients who received no radiotherapy (RT), those who received a mean pancreas dose < 2,300 cGy, and those who received a mean pancreas dose ≥ 2,300 cGy.
Across all patients, baseline HbA1c levels were similar between RT and no RT groups (p=0.159). The mean 2-year elevation in HbA1c levels was higher in the RT group (0.15±0.19 vs. 0.06±0.22, p=0.040), though the 2-year HbA1c levels between the two groups did not differ (5.61±0.39 vs. 5.60±0.35, p=0.886).
Stratification by 2,300 cGy showed comparable baseline HbA1c levels (5.49±0.32 in no RT/< 2,300 cGy group vs. 5.49±0.52 in ≥ 2,300 cGy group, p=0.964). The mean 2-year elevation in HbA1c levels was significantly higher in the ≥ 2,300 cGy group (0.31±0.20 vs. 0.09±0.20, p=0.001). Additionally, the 2-year HbA1c level was significantly higher in the ≥ 2,300 cGy group (5.80±0.45) compared to the no RT/<2,300 cGy group (5.58±0.34, p=0.042).
We performed univariate analysis to identify other risk factors for a 2-year elevation in HbA1c levels (Table 2). However, except for a mean pancreas dose ≥ 2,300 cGy, all other factors, including age, sex, BMI, and baseline HbA1c level, had p-values > 0.100, so multivariate analysis was not considered.
Comparison of the 2-year elevation in HbA1c levels using the Mann-Whitney U test (n=103)
3. Time to event analysis
The median follow-up period for HbA1c levels was 37 months (IQR, 25 to 75 months). During the follow-up, an elevation in HbA1c level of ≥ 0.5% occurred in 17 patients with median time to event of 33 months (IQR, 20 to 53 months). Cox regression analysis identified a mean pancreas dose ≥ 2,300 cGy as a significant predictor of HbA1c elevation ≥ 0.5% in both univariate and multivariate analyses (p < 0.001) (Table 3). Consistent with 2-year HbA1c findings, patients receiving a mean pancreas dose ≥ 2,300 cGy had a significantly higher risk than no RT (p < 0.001), whereas patients receiving < 2,300 cGy did not show a significant difference from the no RT group (p=0.851). The estimated 3-year HbA1c elevation of ≥ 0.5% rates were 5.2%, 8.0%, and 35.2% in no RT, < 2,300, and ≥ 2,300 cGy, respectively (p < 0.001) (Fig. 3). Obese patients (BMI ≥ 25 kg/m2) had significantly higher risk in univariate analysis (p=0.030) but not statistically significant in multivariate analysis (p=0.053). No significant associations were found for the other factors.
Univariate and multivariate Cox regression analyses of risk factors for an elevation in HbA1c level of ≥ 0.5% (n=103)
Kaplan-Meier curves for time to hemoglobin A1c (HbA1c) elevation ≥ 0.5% in patients who received no radiotherapy (RT), those with a mean pancreas dose < 2,300 cGy, and those with a mean pancreas dose ≥ 2,300 cGy (p < 0.001).
Six of 103 patients (5.8%) developed DM, with a median onset of 41 months (IQR, 24 to 77). No cases occurred in the <2,300 cGy group, so they were combined with no RT. Baseline HbA1c ≥ 5.7% (prediabetes) (p=0.023) and mean pancreas dose ≥ 2,300 cGy (p=0.017) were significant DM risk factors in the multivariate analysis (S3 Table).
Among 27 patients with baseline HbA1c ≥ 5.7%, three of five (60.0%) in the ≥ 2,300 cGy group developed DM vs. two of 22 (9.1%) in no RT/< 2,300 cGy. Among 76 patients with HbA1c < 5.7%, only one developed DM after 6 years. The estimated 3-year DM rate was 20.0% in patients with both baseline HbA1c ≥ 5.7% and pancreas dose ≥ 2,300 cGy, compared to 1.4% in others (p < 0.001).
4. Oncological outcome
Irrespective of whether they received RT, all patients achieved complete remission of lymphoma, with no recurrence during the follow-up period.
Discussion
In this study, we identified a dose-response relationship between mean pancreas dose and the 2-year elevation in HbA1c levels. Notably, when stratified using a mean pancreas dose threshold of 2,300 cGy, patients who received ≥ 2,300 cGy had a significantly higher risk of HbA1c elevation compared to those who received no RT or < 2,300 cGy, whereas no significant difference was observed between the no RT and < 2,300 cGy groups. Consequently, patients in the ≥ 2,300 cGy group had significantly higher 2-year HbA1c levels than those in the no RT/< 2,300 cGy group. Similarly, a mean pancreas dose of ≥ 2,300 cGy remained a significant risk factor for both an HbA1c elevation of ≥ 0.5% and the development of DM in time-to-event analysis. These findings suggest that a mean pancreas dose of ≥ 2,300 cGy may increase the risk of an elevation in HbA1c levels, whereas doses < 2,300 cGy may have a limited impact in older adults.
Unlike the liver and kidneys, the pancreas is not traditionally considered a critical OAR during RT [11,12]. Research involving children and young patients who received total body, abdominal, or para-aortic irradiations has suggested an association between pancreatic irradiation and increased DM risk in the long term [13-16]. We previously demonstrated that older adults receiving RT for indolent gastroduodenal lymphomas exhibited elevated HbA1c levels and an increased DM risk compared with non-irradiated patients, implicating pancreatic irradiation as a potential risk factor [22]. Furthermore, a difference in DM risk between RT and non-RT groups using a mean pancreas dose threshold of 2,100 cGy was noted in a later study in another institution involving patients with gastric MALToma [23].
High HbA1c levels are associated with an increased DM risk [17,18] and are also linked to a higher risk of cardiovascular events, even in non-diabetic individuals [19,20]. We previously demonstrated that patients who received RT had a higher risk of HbA1c elevation than those who did not receive RT; however, our study was limited by a small sample size and undefined testing interval. In the current study, we maintained HbA1c as the primary measure, analyzed a larger number of patients, and compared HbA1c levels based on the mean pancreas dose at a 2-year follow-up point. Our findings confirmed that compared with the other two groups, the group with a mean pancreas dose of ≥ 2,300 cGy revealed a significantly higher risk of HbA1c elevation. No significant difference was observed between the patients who did not receive RT and those receiving a mean pancreas dose of < 2,300 cGy.
Traditionally, the pancreas has not been considered an OAR during RT. However, our study showed that pancreatic irradiation significantly increases HbA1c level. Notably, we also observed that when pancreatic irradiation doses are sufficiently minimized, the HbA1c elevation becomes comparable to that of non-irradiated patients. These findings highlight the importance of considering the pancreas as a potential OAR in RT planning. This is particularly crucial for patients with prediabetes (HbA1c level ≥ 5.7%), who are at increased risk of developing DM.
Techniques such as respiratory control, intensity-modulated RT, and proton therapy can be beneficial [23]; however, dose reduction is also one of the most effective strategies. Historically, gastric MALTomas have been treated with doses exceeding 30 Gy [2,9]. However, recent evidence reveals a comparable complete response rate with a reduced dose of 24 Gy, leading to increased adoption of the 24 Gy regimen [7,9]. Based on the evidence for reduced-dose RT and suggested association between pancreatic irradiation and DM risk, our institution modified the gastric MALToma regimen from 30 Gy in 15 fractions to 24 Gy in 12 fractions in 2020, resulting in a corresponding reduction in the mean pancreas dose, as demonstrated in this study. Additionally, a recent pilot study proposed that a 4-Gy initial dose followed by a 20-Gy boost for patients with residual diseases could be considered in future treatment protocols [24].
Despite these findings, several limitations must be acknowledged. As a single-institution retrospective analysis, the generalizability of our findings is inherently limited. Additionally, given the rarity of this patient population, the sample size was constrained. Although we conducted both a 2-year analysis and a time-to-event analysis, the median HbA1c follow-up period was 37 months. Furthermore, while baseline characteristics were generally balanced between the RT and no RT groups, patients who received 30 Gy showed a non-significant tendency toward higher baseline HbA1c levels compared to those who received 24 Gy. Minor differences in other metabolic risk factors such as BMI and hypertension were also noted. These potential imbalances may have introduced residual confounding and should be considered when interpreting the dose–response relationship. These limitations may have constrained our ability to draw definitive conclusions, particularly for doses below 2,300 cGy. To address these limitations, we are currently conducting a prospective trial (IRB number: 2021-02-116) to evaluate HbA1c elevation and DM risk in patients with indolent gastroduodenal lymphoma treated with RT. Moving forward, multi-institutional and prospective studies involving larger cohorts will be essential to validate our results further and optimize radiation planning strategies.
Overall, our findings demonstrate that pancreatic irradiation dose significantly influences HbA1c elevation and subsequent DM risk in older patients with indolent gastroduodenal lymphoma. Specifically, a mean pancreas dose of ≥ 2,300 cGy was strongly associated with an increased risk of HbA1c elevation, whereas lower doses showed minimal effects. Therefore, strategies to minimize pancreatic irradiation through dose reduction or advanced RT techniques should be considered in clinical practice to mitigate potential long-term complications.
Electronic Supplementary Material
Supplementary materials are available at Cancer Research and Treatment website (https://www.e-crt.org).
Notes
Ethical Statement
This study was approved by the institutional review board (IRB) of Samsung Medical Center (approval number: 2024-09-098) and conducted in accordance with the ethical principles outlined in the Declaration of Helsinki. The Institutional Review Board waived the requirement for signed informed consent from patients or their guardians due to the retrospective nature of the study.
Author Contributions
Conceived and designed the analysis: Baek JY, Lim DH.
Collected the data: Baek JY, Lim DH, Oh D, Kim JJ, Lee JH, Min BH, Lee H.
Contributed data or analysis tools: Baek JY, Lim DH, Oh D.
Performed the analysis: Baek JY, Lim DH.
Wrote the paper: Baek JY, Lim DH, Oh D, Kim JJ, Lee JH, Min BH, Lee H.
Conflicts of Interest
Conflict of interest relevant to this article was not reported.