Widespread Multimorbidity in a Cohort of Aging, Radiation-exposed Rhesus Macaques.

The cohort for these analyses consisted of a 277-animal subset in the RLEC as of June 2022 (170 living and 107 deceased) including all diagnoses made since January 2012 when regularized imaging began in the cohort leading to an increase in diagnoses. Of these, 226 received whole-body ionizing irradiation (67 females) and 51 are non-irradiated control animals (4 females). Due to the history of male animals predominantly being used for radiation experiments, there were fewer females in the colony overall with only 13 deceased females in the sample. While there was no difference in age at death between the males and females, living females were 9.21 years old on average compared to 10.3 years in living males (Welch’s t-test P = 0.04) because of the historical bias for males. This may impact associations between age of onset and sex; therefore, we focused on the overall relationship between irradiation, aging, and disease onset including both sexes in a single analysis. Subsequent work will be able to evaluate sex-specific effects in the cohort. Animals were exposed to 4.0–8.5 Gy of ionizing radiation (mean 6.17 ± 1.29 Gy) between the ages of 2 and 10 years (mean 4.6 ± 1.5 years). Figure 1 describes the distribution of living and deceased, irradiated and control animals in this study by age and sex. As additional animals have been added to the cohort, there is a broad range in years of follow-up since irradiation ranging from less than one year to more than 15 years with a mean of 5.2 ± 3.0 years. Deceased animals were evaluated for an average of 6.4 ± 3.6 years postirradiation. All in vivo blood collections and other postirradiation procedures were conducted at the Wake Forest University School of Medicine with approval by the Institutional Animal Care and Use Committee of Wake Forest University. Wake Forest University has an Assurance on file in the Office for Protection from Research Risks, Office of the Director, National Institutes of Health, that accepts responsibility for the humane care and use of animals (OPRR #A-3391-01). The Laboratory Animal Care Program of the Wake Forest University School of Medicine complies with the Principles for Use of Animals, the Guide for the Care and Use of Laboratory Animals ( 20 ), all provisions of the Animal Welfare Act, and has been accredited by the Association for Assessment and Accreditation of Laboratory Animal Care, International (AAALAC) since April 8, 1966 (AAALAC File #8). The control and irradiated animals used for this study were obtained from Wake Forest University, University of Maryland, University of Illinois, Armed Forces Radiobiology Research Institute, Lovelace Respiratory Research Institute, MD Anderson, National Institutes of Health, Citox Labs, Worldwide Primates Incorporated, and Primate Products. Animals received total-body irradiation under IACUC oversight at their prior institution using one of two strategies: 1. linear accelerator-derived photons at a nominal mean energy of 2 MeV, delivered at 80 cGy/min as a split dose given half anterior-posterior and half posterior-anterior; or 2. Cobalt 60-derived gamma radiation delivered simultaneously, bilaterally at 60 cGy/min. These are potentially lethal doses as the LD 10/30 for rhesus macaques is ~5.5 Gy, the LD5 0/30 is ~6.7 Gy, and the LD 90/30 is 8 Gy ( 21 ). Supplementary Fig. S1 2 ( https://doi.org/10.1667/RADE-24-00014.1.S1 ) shows the distribution of radiation doses in the study sample. Although acute responses to radiation occurred prior to most animals arriving at Wake Forest, 156 animals were exposed to radiation doses overlapping the reported dose range for the hematopoietic acute radiation syndrome (H-ARS, 6.25 Gy–8.75 Gy) ( 22 ) with 45 animals exposed to greater than 7 Gy. Animals surviving H-ARS were subsequently transferred to Wake Forest University School of Medicine Center for Comparative Medicine Research for long-term monitoring postirradiation. Irradiation methods, supportive care strategies, and acute effects for many animals donated to this cohort have been reported ( 21 , 23 , 24 ). Irradiated animals arrived at Wake Forest as few as 61 days postirradiation and as long as 4.5 years with a median of 203 days. Animals were fed the Typical American Primate diet (LabDiet 5L0P; Land O’Lakes Inc., St. Louis, MO) designed to approximate the macronutrients of a Western dietary profile. This diet was supplemented with fresh fruits and vegetables and with water ad libitum. Eighty-five percent of the animals were housed socially in indoor-outdoor pens, while 15% were housed in group cages for safe handling or medical care. Care was taken to ensure the animals in the groups were compatible. Environmental enrichment, including fruits/vegetables, toys, puzzles, climbing and hiding environments, was provided continuously on a rotating basis. Behavioral well-being was additionally monitored, and recommendations were made as needed by an independent behavioral management team. All animals were trained to cooperate in handling procedures, to minimize stress. Sampling was scheduled so that the animals were sedated the minimum number of times required for data collection. All animals dying (11%) or euthanasia (89%) in the cohort underwent a complete post-mortem examination, including gross and histopathologic examination of all organ systems, and collection of archival tissue samples. Additional details about the management, demographics, and husbandry of the RLEC are presented by Johnson et al. ( 25 ). After arrival at Wake Forest, all animals were monitored twice daily by trained veterinary technical laboratory staff to assure animal well-being and social stability. Animals are observed for health and behavioral issues daily with clinical blood chemistries and complete blood counts three times per year during routine clinical exams. Any evidence of illness was classified using a nonhuman primate-specific modification of the Children’s Clinical Oncology Group toxicity criteria ( 26 ). Sick animals were promptly evaluated by experienced veterinary staff, board-certified by the American College of Laboratory Animal Medicine. Diagnoses are made either by physical exam or blood biomarkers during these exams or during annual abdominal ultrasounds, dual X-ray absorptiometry (DEXA), full-body computerized tomography (CT) scans, echocardiograms (ECG), magnetic resonance imaging (MRI), or gastrointestinal endoscopy. Diagnostic criteria for each chronic disease or group of diseases (e.g., heart disease refers to any cardiovascular disorder) analyzed in this paper are described in Supplementary Table S1 ( https://doi.org/10.1667/RADE-24-00014.1.S1 ) and detailed postmortem findings including sub-classifications of disease are described by Schaaf et al. ( 27 ). For evaluating overall and cumulative incidences of disease, we consider an animal that has ever been diagnosed with one of these chronic conditions to have that disease through the end of life, even if the symptoms resolve. This is consistent with diagnostic practices (e.g., “problem lists”) from human health records. The primary outcomes are lifespan and healthspan, which is defined here as time to first chronic disease diagnosis. Kaplan-Meier curves to compare lifespan and healthspan between irradiated and control animals were calculated using the survival package in R ( 28 , 29 ). Welch’s t-test was used to evaluate mean differences in lifespan between deceased irradiated and control animals. This test was selected due to the differences in sample size and variance in age at death between the two groups. To test for dose effects on lifespan and healthspan, we fit a linear regression model. Welch’s t-test was used to test for significant differences in mean age at diagnosis for each disease in all affected animals, living or deceased. To evaluate the impact of radiation on lifetime disease incidence, two-sided Pearson’s chi-squared tests determined the difference in proportion of affected animals between deceased, irradiated and control animals. Proportions were considered significantly different if the 80% confidence interval did not overlap zero. This relatively liberal cutoff value was chosen due to the relatively small sample size of 20 deceased, control animals. All analyses were performed in R version 4.2.1 ( 30 ). Welch’s t-tests were used to evaluate differences in the total number of diagnoses between irradiated and control animals at each age from the beginning of adolescence at age 4 to age 16, the oldest age to which at least 10 irradiated and control animals survived. Multiple morbidity values were quantified by a simple count of diagnoses up until that age regardless of whether symptoms resolved at any point prior to or after the index age. To identify diseases that co-occur in the same animal more often than expected by chance, we considered all irradiated animals in the cohort, living and deceased. Using chi-square goodness of fit test, we compared the observed number of animals in which each of the 190 pairs of diseases is comorbid to the expected number of animals in which the diseases would be comorbid if the diseases were distributed randomly with the observed frequency. Diseases were considered significantly comorbid based on a Bonferroni-corrected α = 2.6 × 10 −4 .

In our study subset, the mean lifespan of irradiated animals is significantly shorter at 11.3 years (standard deviation ± 4.0) on average compared to 16.4 years (±3.57) in control animals (Welch’s t-test P = 3.2 × 10 −6 , Kaplan-Meier log rank P = 2.8 × 10 −4 ). While radiation has a clear impact on the lifespan, there is no evidence of an effect of radiation dose on after exposure lifespan in this cohort once control animals (0 Gy) are removed [R 2 = 0.02, P = 0.11, β 95% confidence interval (−0.81, 1.037)]. Comparison of the mean lifespan between the animals with the lowest exposures (≤5 Gy) and the highest exposures (≥8 Gy) show a non-significant difference of 2.4 years. Given the wide variance in lifespan in these groups, a post hoc power calculation indicates the true difference in mean lifespan between the animals at the extremes of our dose distributions is less than 4 years. To evaluate the impact of radiation on healthspan, the time to first diagnosis was calculated for each animal, excluding periodontitis due to its extremely high prevalence. The average age at first diagnosis of any disease was 6.8 in irradiated animals and 10.5 in controls (Welch’s t-test for mean difference P = 2.1 × 10 −6 , Kaplan-Meier log rank P = 3.3 × 10 −8 ). Dose effects were non-significant [R 2 = −0.006, P = 0.65, β 95% confidence interval (−0.30, 0.39)]. When only the highest and lowest exposure groups are compared, high-dose exposed animals are diagnosed with their first disease 4.0 years younger than those exposed to the lowest dose (Welch’s t-test P = 0.0004). Figure 2 shows the distributions of lifespan and healthspan in this cohort. Figure 3 shows the total lifetime incidence for each disease among the 107 deceased animals by radiation status. In living animals, disease incidence is lower overall as unaffected younger animals may develop disease later in life. Most diseases have higher incidence among irradiated animals with the proportion of affected irradiated animals being statistically significantly higher for cataracts, testicular atrophy, tumors, diabetes, and brain lesions based on an 80% confidence interval cutoff for the chi-squared test ( Supplementary Table S2 ; https://doi.org/10.1667/RADE-24-00014.1.S1 ). For several diseases, incidence is lower in irradiated animals compared to control animals; however, this difference is only statistically significant for periodontitis and being overweight. Given the obesogenic diet on which the RLEC is maintained, being of normal weight could be considered a metabolically abnormal condition. Not only is there a difference in disease incidence between irradiated and control macaques, but for many diseases there is a difference in age of onset as well. Figure 4 shows the age of diagnosis by radiation status for all affected animals in the full 277 animal dataset. The vast majority of diagnoses are made during the lifetimes of the animals with the exceptions of 58% of tumors, 13% of heart disease, 5% of kidney disease, and 2% of lung disease which are diagnosed at necropsy. Therefore, age of onset may be upwardly skewed for tumors and heart disease to a greater degree than for other diseases. For every disease except gastrointestinal disorders (most commonly diarrhea), irradiated animals are diagnosed at a younger age than control animals. This difference is statistically significant based on Welch’s t-tests for periodontitis, cataracts, bone mineral density, overweight status, and arthritis (full results shown in Supplementary Table S3 ; https://doi.org/10.1667/RADE-24-00014.1.S1 ). Of 107 deceased animals, eleven died without any diagnoses. All were irradiated and their mean age at death was 8.6 years, nearly three years younger than the average irradiated animal. There are currently 69 living animals without any diagnoses but most of them are young with a mean age of 7.38. Approximately 25% of animals that live to age nine have zero diagnosed disorders (42/166 animals). The oldest animal ever to have zero diagnoses died at age 12.1, although there is a currently living 11.9-year-old also with zero diagnoses. Beginning at age 8, irradiated animals have more accumulated diagnoses than control animals with significant differences (P < 0.05) for all ages 8–16 except age 14. Figure 5 shows the mean differences in morbidity counts between control and irradiated animals and highlights the broader distribution at all ages among the irradiated animals. There is a weak correlation between radiation dose and number of comorbid conditions among the irradiated animals (linear adjusted R 2 = 0.01, P = 2.0 × 10 −4 ). The emergence of multiple chronic conditions in a given animal is not randomly distributed, but clusters as shown in Fig. 6 . Table 1 shows all pairs of diseases co-occurring more often than expected by chance based on a chi-squared test (Bonferroni corrected P < 0.05, full pair-wise comparisons in Supplementary Table S4 ; https://doi.org/10.1667/RADE-24-00014.1.S1 ). Diabetes and cataracts are major hubs, each showing excess comorbidity with four other diseases. Bone mineral density, periodontitis, and heart, kidney, and lung diseases all have three linked diseases, while hepatic dysfunction, testicular atrophy, and tumors each have two.

The RLEC is unique in the extended period of monitoring after irradiation as well as the depth and frequency of imaging and other diagnostic tools employed to evaluate the health status of the animals. This enables us to compare both the incidence and age of onset of chronic conditions with a high degree of accuracy. With the exception of testicular atrophy, all of the chronic diseases observed among RLEC animals develop naturally among aging rhesus macaques, although not all are seen in the controls for the RLEC cohort. However, all-cause mortality and morbidity are higher in radiation-exposed animals relative to controls with several previously reported diseases having statistically significantly greater lifetime incidence among irradiated animals compared to controls, for example cataracts ( 31 ), testicular atrophy ( 15 ), tumors ( 18 ), diabetes ( 3 ), normal weight (on an obesogenic diet), and brain lesions ( 5 ), and many other diseases showing non-significant, but increased incidence. Notably, being overweight is more common in control animals. This may be associated with adipocyte dysfunction in irradiated animals that results in decreased fat mass relative to controls ( 2 ). We have calculated lifetime incidence only in the deceased animals (67 irradiated, 20 control) because the distribution of ages in living control and irradiated animals is skewed which could lead to inaccurate results. As more animals are tracked through the end of their lives, we will be able to power more robust comparisons of disease incidence, particularly for less common diseases. Similarly, because the 50% of animals were exposed to between 6 and 7 Gy, the ability to resolve dose-dependent effects, particularly among the less common diseases, is limited until more cases have been observed. Age of onset among affected irradiated animals was younger than for affected control animals for all diseases except gastrointestinal disorders and significantly different for periodontitis, cataracts, bone density, being overweight, and arthritis. For some of these disorders, particularly bone loss and arthritis, diseases that would typically appear late in life (if at all) in control animals appear by middle age in irradiated animals. Bone loss, as an example, appears at age 16 on average in the control animals reflecting approximately 66% of the maximum lifespan seen in our cohort. In irradiated animals, the onset of bone loss occurs more than seven years earlier on average at age 8.7 (36% of maximum lifespan). Among non-significantly different diseases, the age of onset is broad and overlapping for both irradiated and control animals, but the trend towards earlier age of onset is consistent. This suggests that across a range of physiologically distinct disease processes, irradiation accelerates progression which contributes to earlier diagnosis and likely to the increased overall incidences of these diseases. Because most of these diseases appear in control animals, we know that they are not exclusively caused by radiation exposure. Rather, the insult to homeostasis caused by radiation increases the likelihood of these diseases developing. Despite the substantial differences in morbidity and mortality between irradiated and control animals, no dose effect was identified. The majority of exposures are in the range of 6–8 Gy, representing 25–90% lethal doses (LD 25 –LD 90 ) of ionizing radiation for rhesus macaques. The few animals exposed to <LD 10 all survived less than 7.5 years postirradiation compared to a mean post-irradiation lifespan for all animals of 6.5 ± 3.6 years, suggesting that even the lower acute exposures may result in shortening of lifespan. Within each exposure level, there is a broad distribution of lifespans. Indeed, the shortest-lived animal (2.75 years post-irradiation) was exposed to 6.77 Gy and the longest-lived animal (23.1 years) was exposed to 6.5 Gy. While radiation contributes to more frequent and earlier onset of disease, the tremendous heterogeneity in the age of onset and number of comorbid conditions is not explained by radiation dose, age at exposure, or sex. This suggests that as yet unidentified underlying differences among the animals contribute to resilience to radiation exposure and subsequent disease processes. These accelerated disease processes may lead to increased total numbers of chronic conditions being diagnosed in each animal with diseases sharing an underlying etiology being comorbid more often than expected by chance. Our results for the number of chronic conditions diagnosed by age in the cohort show significant divergence between irradiated and control animals beginning in early adulthood (age 8). While the mean morbidity counts diverge between irradiated and control animals, the range in morbidity counts also increases markedly with age for the controls and, to a greater extent, among the irradiated animals. Some irradiated animals live well into adulthood (as long as age 12) without ever being diagnosed with a disease. Again, this heterogeneity suggests underlying variability in resilience to the physiological challenges of both irradiation and aging. When evaluating patterns of disease and resilience in the RLEC, it is important to remember the inherent survival bias in the sample. Only animals that have survived the initial acute exposure to ionizing radiation and unirradiated controls are represented in the RLEC and given the range of ionizing radiation exposures, we expect more than 90% to less than 10% survival in the short term depending on the study. Thus, we may be seeing a paradoxical relationship where the animals present in the RLEC who survived the highest exposures to radiation had the most intrinsic resilience leading them to be robust to disease throughout life, while animals exposed to lower doses of radiation may or may not have survived higher doses and could be biologically more at risk of developing chronic disease. Furthermore, the least resilient animals, those that did not survive the acute radiation exposure, are absent from the sample, and they may have been the animals most at risk of developing chronic diseases as they aged. Surviving the initial acute ionizing radiation challenge and the protracted challenge of aging suggests greater levels of resilience. Indeed, aging long enough to develop multiple chronic diseases may, in and of itself, be a biomarker of resilience. This could explain, in part, the lack of a dose-response relationship between the amount of ionizing radiation and postirradiation lifespan and healthspan. It is also consistent with the observation that all the animals in the cohort that have died without any diagnosed diseases were irradiated and died nearly three years younger than the average irradiated animal. The most intrinsically resilient animals (e.g., those most likely to live longer regardless of environment) are also most likely to survive higher doses of radiation and to live long enough to acquire multiple chronic conditions. This pattern is likely to mirror that seen among humans exposed to low- or high-dose radiation. Within these patterns of increased total multimorbidity among the irradiated animals, there are clusters of specific disorders that are frequently comorbid. The network graph of significantly comorbid conditions reveals three main clusters ( Fig. 6 ). The largest has two hubs, diabetes and cataracts, with multiple connections among heart disease, tumors, and lung, kidney, and liver dysfunction. This large cluster of comorbid conditions reflects the multi-organ involvement commonly seen among the delayed effects of acute radiation exposure (DEARE) ( 9 , 10 ). The single strongest association is between diabetes and being overweight, which are comorbid for more than half of diagnoses. However, many diabetic animals are never overweight, in keeping with prior findings that diabetes in irradiated NHPs can arise without the development of adiposity ( 3 ). In these animals, diabetes may be driven by the insufficient response to insulin production by fibrotic skeletal muscle ( 7 ). In the RLEC, systemic inflammation and/or tissue-specific increases in fibrosis has also been linked to cancers ( 18 ), myocarditis ( 1 ), nephropathy ( 32 , 33 ), lung injury and pneumonitis ( 34 – 36 ). Other NHP models of radiation injury have also identified fibrosis in the liver ( 37 ) and increased risk of cataracts among radiation-exposed humans may be mediated by pro-fibrotic pathways ( 38 , 39 ). Fibrosis is visible histologically in RLEC animals across a range of tissues ( 27 ) and transcriptional profiling of peripheral blood monocytes has shown systemic pro-fibrotic changes in the months after irradiation ( 40 ). The relatively late onset of fibrosis-associated diseases in the RLEC animals relative to the timing of irradiation suggests that the acute tissue damage due to radiation exposure is compounded by chronic inflammation and aging that exacerbates the accumulation of fibrotic proteins until the body is unable to efficiently clear the extracellular matrix components, leading to organ dysfunction. Other prominent disease clusters are centered on bone loss and periodontitis. Bone loss is a significant and relatively rapid result of radiation ( 8 ), and frequently comorbid with being underweight and female reproductive disorders, a finding that persists when only female animals are considered. These findings are consistent with human studies of older adults given the mechanistic relationship between weight bearing and bone remodeling, and underweight linked to dysregulation of gonadotropins, a known result of radiation therapy in human young adults ( 41 ). Cataracts are linked to periodontitis which forms the final cluster with skin, testicular atrophy, and hypertension. Recent papers have demonstrated an irradiation dose-response relationship for hypertension and testicular atrophy, in contrast with other disorders that do not show a dose-response relationship. This may lead to animals with similar radiation exposures having the same risk of developing these conditions ( 15 , 42 ). In the RLEC, these comorbidities may reflect not a specific underlying biological pathway, but a broader inherent resilience and ability or inability to maintain homeostasis within organ systems. Overall, these findings highlight the profound impact of radiation across organ systems leading to the accelerated development of chronic diseases of aging among irradiated animals. This is true both for individual diseases and for total morbidity, which is greater among irradiated animals compared to controls. Radiation-associated multimorbidity may be driven by disequilibrium in biological processes leading to systemic increases in fibrosis among other shared etiologies. Indeed, the buffering capacity of an individual or an organ system in handling the disruption of radiation may explain inter-individual and inter-disease differences in the latent period before clinically detectable manifestations of disease. While the widespread multimorbidity among irradiated animals reflects the profound, acute challenge to homeostasis, the heterogeneity within animals exposed to the same doses of ionizing radiation suggests a residual, but as yet unidentified, biological source of resilience that differs across the animals.

The development of multiple chronic conditions, or multimorbidity, can result from the erosion of biological resilience, the ability to recover from deviations from the homeostatic state. Long-term-monitoring of common, chronic diseases in the Wake Forest Non-Human Primate (NHP) Radiation Late Effects Cohort (RLEC) has led to the identification of 20 diseases occurring in at least 5% of irradiated and/or control rhesus macaques ( Macaca mulatta ) over 3–7 years postirradiation. Many of these individual chronic diseases have been described in the RLEC previously, including myocardial fibrosis and systemic inflammation ( 1 ), type 2 diabetes and insulin resistance in adipose tissue ( 2 , 3 ) white-matter injury ( 4 ), cerebrovascular lesions ( 5 ) and cognitive impairment ( 6 ), skeletal muscle fibrosis ( 7 ), and cortical thinning and structural changes in bone ( 8 ). In humans and model organisms including NHPs and mice, these conditions are frequently comorbid, composing a suite of conditions known as delayed effects of acute radiation exposure (DEARE) ( 9 – 13 ). Multimorbidity is also common among older adults and NHPs diagnosed with hypertension, diabetes, heart disease, arthritis, pulmonary diseases, and/or cognitive impairments. Two of the most common sequelae of radiation exposure in NHPs are cataracts and testicular atrophy which are deterministic effects with previously established threshold doses of approximately 4 Gy and 20 cGy of radiation, respectively, and a dose-dependent increase in frequency above this threshold ( 14 – 16 ). In contrast, stochastic sequelae show a dose-dependent increase in the likelihood of diagnosis beginning at low doses. Cancers are common stochastic effects in both NHPs and humans ( 17 , 18 ). Differentiating between deterministic and stochastic patterns for late effects is particularly challenging both because the prevalence of these diseases is low both in irradiated and control animals and because the baseline likelihood of diagnosis increases with age. The interaction between the acute, early-life stressor of irradiation and the ongoing, systemic stressor of aging jointly contribute to the degradation of resilience which can lead to the accumulation of chronic diseases in radiation-exposed animals that resemble humans of a much more advanced age ( 19 ). To understand the distribution of these individual chronic diseases as well as patterns of comorbidity across diseases, we have analyzed a decade of clinical records in the RLEC. With these data, we can determine the combined influence of radiation and aging on the incidence and age of onset for each disease in isolation as well as the disease burden across the postirradiation lifespan. The depth of imaging and pathology allows for granular identification of multimorbidity in each animal across its life course to identify patterns of comorbidity that reflect shared radiation- and aging-driven etiologies.

Supplementary Fig. S1. Histogram of radiation doses for all animals in study sample with lethal dose (LD) range indicated. Supplementary Table S1. Diagnostic criteria for the diseases included in this analysis. Supplementary Table S2. Results of two-sided Pearson’s chi-squared tests test for difference in proportion of affected deceased controls and affected deceased irradiated animals. Supplementary Table S3. Results for Welch’s t-test comparing mean age of onset for each disease between affected irradiated and control animals. Supplementary Table S4. Results of χ 2 test comparing observed incidence of pairwise comorbidity to expected incidence based on frequency of disease diagnosis in the cohort.