Synopsis of Partial-Body Radiation Diagnostic Biomarkers and Medical Management of Radiation Injury Workshop

Abstract

Radiation exposures from accidents, nuclear detonations or terrorist incidents are unlikely to be homogeneous; however, current biodosimetric approaches are developed and validated primarily in whole-body irradiation models. A workshop was held at the Armed Forces Radiobiology Research Institute in May 2008 to draw attention to the need for partial-body biodosimetry, to discuss current knowledge, and to identify the gaps to be filled. A panel of international experts and the workshop attendees discussed the requirements and concepts for a path forward. This report addresses eight key areas identified by the Workshop Program Committee for future focus: (1) improved cytogenetics, (2) clinical signs and symptoms, (3) cutaneous bioindicators, (4) organ-specific biomarkers, (5) biophysical markers of dose, (6) integrated diagnostic approaches, (7) confounding factors, and (8) requirements for post-event medical follow-up. For each area, the status, advantages and limitations of existing approaches and suggestions for new directions are presented.

INTRODUCTION

Although most biodosimetry research focuses on whole-body irradiation, a review of radiation accidents shows that the majority of cases are non-homogeneous exposures. For example, in a criticality accident in Tokaimura, Japan (September 1999), three men received very high doses of c and neutron radiation. Application of a multiple parameter diagnostic biodosimetry approach in the medical management of these radiation accident victims was reviewed (1). Dicentric analysis of circulating lymphocytes reflected a non-homogeneous exposure (2), most likely because of partial shielding at the time of the exposure and also because of neutron attenuation in tissue. The clinical course of one of the patients reflected this inhomogeneity; despite whole-body equivalent doses greater than 8 Gy, the patient showed recovery of bone marrow cells (3). Inadvertent handling of radioactive materials is another avenue for partial-body radiation exposures. In Thailand (February 2000, Samut Prakaran, Bangkok) several people were accidentally exposed to ⁶⁰Co γ rays from a partially dismantled Gammatron teletherapy source while recycling scrap metal. Those who procured the source, junkyard workers and relatives of these workers received the largest radiation doses. Some scrap metal collectors received severe localized cutaneous radiation injuries; however, their whole-body doses were only around 2 Gy (4). Reconstructed doses using the rapid interphase chromosome aberration (RICA) assay (5) correlated well with peripheral blood lymphocyte count decreases and severity of clinical symptoms of radiation sickness in nine individuals (6).

In such accidental radiation over exposure events, the inhomogeneity of exposure must be considered before treating acute radiation syndrome (ARS); non-homogeneous exposures may require treatments different from total-body exposures for the same total dose. For example, if there is bone marrow sparing, despite high doses, bone marrow transplants would not be appropriate; instead cytokine therapy may be considered (7, 8). Rapid diagnosis of organs at risk after a non-homogeneous exposure would allow targeted treatments.

In May 2008, the Armed Forces Radiobiology Research Institute (AFRRI) held a Partial-body Radiation Diagnostic Biomarkers and Medical Management of Radiation Injury (PB-RAD-Injury) workshop to (1) discuss the current state of knowledge about partial-body exposures, (2) identify current approaches and available biomarkers, (3) review the impact of partial-body biodosimetry on treatment, (4) identify research gaps, and (5) propose future directions. Seventeen experts presented reviews of the state of the science in a variety of areas. Extended abstracts for the Workshop speakers and 12 poster presenters are accessible on AFRRI’s website (http://www.afrri.usuhs.mil/outreach/reports/pdf/sp09-1.pdf) and as Supplementary Information. Approximately 90 scientists from around the world attended the workshop.

The meeting opened with a review of both partial-body radiation accidents and use of cytogenetic and molecular biomarkers for assessment of inhomogenous exposures using in vitro and animal models of partial-body exposures. Workshop sessions were organized by physiological system [cutaneous system (skin), hematopoietic system (bone marrow and blood), gastrointestinal system (gut), and other organs (i.e., lung, kidney, liver)] and focused on the current status of diagnostic approaches to assess radiation injury or dose. An additional session was devoted to exploring novel approaches (i.e., molecular biomarkers, biophysical dosimetry) for partial-body biodosimetry. The final lectures described current approaches for planning and medical treatment of partial-body exposures. The presentations provided the framework for a discussion that addressed the current directions, research gaps, plans for overall integration of dose assessment approaches, and clinical applications of the technologies presented. For this report, highlights were culled from the talks and discussions at the workshop. This report is not intended to be an exhaustive review of the field but rather a summary of the workshop discussions on the identified focus areas.

WORKSHOP FOCUS AREAS

Cytogenetic Analysis

Cytogenetics is an established approach for the assessment of radiation exposure. The dicentric bioassay in mitogen-stimulated blood lymphocytes is considered the gold standard (9, 10). In the typical application of the metaphase-spread dicentric bioassay, the mean number of dicentrics per cell measured for a suspected radiation-exposed individual compared to an appropriate in vitro radiation dose–response calibration curve provides an approximate total-body dose estimate. The same dicentric metaphase-spread bioassay, with an additional analysis of the frequency distribution of chromosomal aberrations including null events (cells without aberrations) and overdispersion (cells with multiple aberrations), can provide insights on the extent and dose for partial-body exposure scenarios (11). Two analytical methods, the contaminated Poisson of Dolphin (12) and the Qdr method of Sasaki and Miyata (13), are typically used to derive the size and dose to the irradiated fraction. Dolphin’s method considers the overdispersed distribution of dicentrics among all analyzed cells. Sasaki’s method considers the yields of dicentrics and rings only in irradiated cells (cells with either a dicentric, ring or fragment) for dose estimation. Both methods provide good estimates of the fraction of the body exposed in cases of prompt (acute) or chronic radiation exposures. These methods, based on use of peripheral blood lymphocytes, cannot specify the exposed body area, but in cases of acute radiation exposures they can provide diagnostic evidence for the sparing of bone marrow, which would affect the therapeutic approach to radiation injury.

Partial-body biodosimetry based on analysis of chromosomal aberrations of circulating peripheral blood lymphocytes makes several assumptions: (a) All of the peripheral blood lymphocytes in the exposed field were exposed to the same radiation dose, (b) the peripheral blood lymphocytes were distributed evenly throughout the body, and (c) the exposure time was sufficiently short to prevent significant mixing of exposed and non-exposed pools. These assumptions are not likely to be strictly met but often are sufficient to provide estimates for exposures. An important caveat for partial-body cytogenetic analysis is that the distribution of hematopoietic bone marrow varies with age, being in the limbs and axial skeleton in children but retreating to the axial skeleton by about age 20 (14, 15). The clinical impact of partial-body exposures will vary with the area of the body exposed and can be expected to be substantially different from a total-body exposure at the identical dose. Classic experiments by Jacobson confirmed this distribution by demonstrating that mice can survive lethal irradiation by shielding of a single limb (16). Use of dicentric analysis, organ-specific protein biomarkers, and peripheral blood gene expression analysis in combination with clinical signs and symptoms may help address these limitations.

Analysis of dicentrics is labor intensive, requires significant expertise to carry out, and takes several days to provide the results. The cytogenetics community is developing strategies to respond to radiological mass casualties, including partial-body exposures. One approach is to score a limited number of metaphases per subject (20 to 50 metaphases) to provide an approximate dose estimate quickly. In vitro simulations have shown that scoring only 50 cells makes it possible to detect 50–95% partial-body irradiations of greater than 2 Gy (17).

Automation of the sample preparation (18) and the image analysis (19) can increase the throughput and reduce the technical skills required to perform the assay. Alternative biomarkers of DNA damage may also be assessed more quickly and easily, providing useful diagnostic information. Examples of potential alternative cytogenetic based radiation bioassays include (a) the RICA assay (5), which measures damage to specific chromosomes in interphase cells by fluorescence in situ hybridization after premature chromosome condensation by mitosis promotion and phosphatase inhibition, (b) the γ-H2AX assay (20), (c) measurement of micronuclei along with other damage in cytokinesis-blocked cells (21), and (d) the premature chromosome condensation assay for high-dose exposure (22). All options and their applicability to partial-body exposure assessment need to be explored.

Clinical Signs and Symptoms

The manifestation of clinical signs and symptoms provides the most important basis for treatment decisions in individual patients (23, 24). Clinical signs and symptoms of radiation exposure also provide diagnostic evidence on the affected organs (25, 26).

Severe and frequent bloody diarrhea clearly indicates gastrointestinal involvement. Neutropenia and thrombocytopenia are definitive indicators of hematopoietic effects. This clinically based approach focuses medical attention on the organs at risk. Some symptoms that have an early onset (e.g. lymphocyte depletion and emesis) are used to provide an indication of the total-body dose and the severity of radiation injury. Many of these symptoms, however, appear with some delay after exposure and perhaps after the injury has progressed significantly. Because of interindividual variability, some symptoms (e.g. skin injury) may not appear (27). Additionally, use of pharmacological countermeasures within days after exposure in exposed individuals might mitigate late-appearing symptoms.

Some clinical symptoms may provide ambiguous information, thus requiring additional biodosimetric approaches to identify concerned public (i.e. worried well), low-dose exposed or life-threatening severely exposed individuals (26). For example, the early erythema after accidental radiation exposure is strong evidence for a high acquired dose, but it may also occur with a purely superficial exposure (e.g. soft X rays, b particles) without bone marrow or other internal effects. Additional assessments, such as blood counts or specific biodosimetry assays, will be needed.

In a radiological mass-casualty situation involving several hundreds of potential victims, following clinical signs for all individuals would be a labor-intensive effort that may not be feasible. The AFRRI Biodosimetry Worksheet (AFRRI Form 331) represents a comprehensive data entry worksheet. It provides a place for recording the facts about a case of radiation exposure, including the source and type of radiation, the extent of exposure, relevant biodosimetry diagnostic information, and the nature of the resulting radiation injuries. The worksheet updated in 2007 now includes a modified version of the METREPOL ARS severity scoring system that was developed by Dr. Fliedner (University of Ulm, Ulm, Germany) and colleagues (23).

Multi-organ interactions are frequently evident in cases of severe life-threatening radiation exposure. Exposure of one organ system may influence the response of another organ system even if it is not in the original field of exposure (28, 29). As we increase our understanding of the underlying mechanisms of radiation injury, indicators of the progression of the syndrome might become more evident. For example, multi-organ failure is likely to be the sequel of the cascade of biochemical events, organ interactions, and vascular and inflammatory effects (30). Understanding the pathophysiology of cellular and organ damage, abscopal effects and systems biology can enhance our understanding of the implications of clinical signs and symptoms.

Cutaneous Bioindicators and Medical Management

Skin provides a unique window on the partial-body distribution of radiation exposures. While it is a complex organ system in its own right, its topography can be exploited for surface demarcation of the radiation exposure pattern. A transient skin erythema can be seen at about 1 h after doses in the range of 3 Gy. Changes in skin such as early and transitory erythema followed by moist desquamation, blistering and later pigmentation and hair loss can give diagnostic information on the topography of partial-body exposures. Changes are graded with dose, but symptoms of radiation-induced skin injuries generally manifest 2–3 weeks after exposure (31). Analyses of radiation accidents reveal that the percentage of affected skin is an excellent prognostic indicator for the severity of multi-organ radiation injury and therefore could be useful for triage and diagnosis.⁶ Immediately after irradiation, skin cells begin to produce cytokines, which catalyze a cascade leading to progressive late symptoms of the cutaneous radiation syndrome (32). Cytokines are important signaling molecules mediating communicative interactions both locally between cell types in dermal tissues and distantly between organs (33). Further work investigating the molecular mechanisms of the cutaneous radiation syndrome is warranted.

Dermal distribution of biomarkers provides another approach to assessment of partial-body exposures. For example, micronuclei in skin fibroblasts are dependent on radiation dose (34). Cytogenetic analysis using skinpunch biopsies at various locations can define the extent of local radiation exposure. A retrospective cytogenetic analysis using excised samples may also be useful for both the study of the pathological course of the cutaneous radiation syndrome and therapeutic choices (35). Alterations in local blood flow after radiation exposure can also be exploited to determine the regional distribution of radiation injury. Techniques in development for this type of approach include optical spectroscopy⁷, ultrasound, Doppler or laser flow profiles⁸, and local temperature changes (32).

Changes in skin and underlying tissues can be used to dictate local treatment of radiation injuries. Using surface changes for assessment of exposure distribution, however, can be seriously confounded by radiation quality. Poorly penetrating radiation may cause severe injury to the skin and immediately surrounding tissues but may not penetrate tissue sufficiently to cause damage to underlying organs. Consequently, the results need to be interpreted carefully.

A new therapeutic approach using local injections of autologous mesenchymal stem cells has shown remarkable efficacy for the treatment of serious cutaneous radiation injuries (36, 37). This novel autograft approach from the French group led by Dr. P. Gourmelon (Institut de Radiprotection et de Surete Nucleaire) has been applied to victims of two radiation accidents (Chile, December 2005; Dakar, June–July 2006). A multiple-parameter and numerical dosimetry approach was used to guide surgical excision of necrotic lesions. Their therapeutic approach was to cover the wound with a skin allograft and then to inject autologous mesenchymal stem cells locally around the lesion. Results showed dramatic efficacy for skin healing as well as immediate but transitory pain relief (38).

Organ-Specific Biomarkers

Early changes in gene expression patterns in peripheral blood lymphocytes can distinguish medically relevant doses of radiation (39, 40) and offer an assessment of overall health status. Studies on the quantitative changes in specific sentinel radiation-responsive genes have led to the use of multiple gene expression targets to formulate consensus dose–response calibration curves for dose assessment applications (39 41). Alternatively, gene expression signatures in isolated cells or tissues can provide organ-specific diagnostic information. MicroRNAs (miRNAs), small non-protein-coding RNAs known to regulate gene expression (42, 43), respond to radiation with tissue and organ specificity. Radiation did not alter the expression of miRNAs in human lymphoblastoid cells (44) but increased expression in lung cancer cells (45). Further studies are clearly needed for an in-depth understanding of the complete dynamics of miRNAome patterns in regulating cellular differentiation, proliferation and apoptosis as well as bystander effects and tissue specificities, which are particularly useful for partial-body exposure scenarios.

Proteomics is another area that offers great potential for early-phase bioindicators of radiation exposure (46–49) that may have applications for partial-body biodosimetry. The ease with which some of these plasma proteomic biomarkers can be measured (46–49) makes this approach attractive for both early- and rapid-diagnostic triage. The use of urinary proteomics offers the benefit of a non-invasive testing that could apply to kidney injury or to whole-body exposures (50, 51).

C-reactive protein (CRP) is an exquisitely sensitive systemic marker of inflammation and tissue damage, and it plays a role in injuries caused by radiation (46, 48, 52–54). The dynamics and level of CRP exactly reflect the time course and severity of the radiation sickness and may be relevant to prognosis (53, 54).

An increase in serum amylase after irradiation of salivary tissue has been proposed as a biochemical dosimeter of early radiation effects (47, 48, 55). The salivary gland has high sensitivity to ionizing radiation (47, 48, 55, 56). Serum amylase level show peak values between 18–30 h after exposure and return to normal levels within a few days. Changes in serum amylase would suggest that the head and neck were irradiated.

Bertho and colleagues proposed plasma flt-3 ligand as a potential bioindicator for radiation-induced aplasia (57) and a correlate of radiation-induced bone marrow damage (58). The number of circulating white blood cells and platelets correlated negatively with plasma flt-3 ligand levels. In a recent radiation accident, the measured flt-3 ligand levels were indicative of the severity of bone marrow aplasia (59).

Plasma citrulline is considered to be a biomarker for small bowel functional epithelial cell mass. Threshold levels predictive for the severity and extent of villous atrophy were identified in patients with villous atrophy disease (60). In a series of experiments and clinical studies, plasma citrulline was validated as a biomarker enabling monitoring of the loss of small bowel functional epithelial cells after ionizing radiation exposure (49, 61–63).

Additional biomarkers are being explored. These include calprotectin, a late marker for gut, fatty acid binding proteins for liver and intestine, glycosphingolipids and their metabolites that might show organ specificity, skin keratins, lung surfactants, oxysterols for organ-specific cholesterol metabolism (59), cytokine cascades, etc.

It is possible that metabolomics can also be used to trace radiation injury to specific organs. Specific changes in these profiles can indicate sentinel targets responsive to radiation injury that may identify the organs at risk. The organ-specific proteomic biomarkers are expressed coincident with the time course of other clinical indicators of organ pathologies, so they provide confirmatory diagnostic information along with other clinical indicators for physicians to develop a science-based medical management treatment strategy. However, these markers could be non-specific, i.e., could also be changed by the effects of trauma, burns or infection, rather than being exclusively the result of radiation (see later section on confounding factors).

The use of biomarkers for the prediction of specific organ damage is appealing but complex in its analysis. Non-specific changes, variability in baseline levels, interindividual variability of the radiation response, and other confounders can complicate interpretation. Changes in one organ may be affected by changes in another physiological system. Specific biomarkers may require invasive sampling (e.g., skin punch biopsies or lung lavage). Biomarkers may also be affected by the health status of the individual and the use of prescription drugs.

Criteria need to be developed to define clinically relevant markers. While many molecules may be modulated by radiation, they may not all be useful for predicting the course of radiation syndromes in patients. Biomarkers are needed to predict both early and late radiation effects. They might also be useful to define the susceptibility of an organ to subsequent insults such as a respiratory infection after radiation-induced lung injury.

Studies on the pathology of organs and mechanistic studies are useful in the exploration of new biomarkers. Understanding the functional implications of the biomarkers will aid in their interpretation and facilitate the search for new indicators. This will allow one to understand not only what is wrong but also why.

Biophysically Based Dose Assessment

Radiation immediately induces stable radicals in teeth and nails that can be monitored by electron paramagnetic resonance (EPR). These signals have the advantage of likely being unaffected by other biological stressors that might appear after a radiological or nuclear exposure. The physical changes are highly reliable, and EPR analysis of dental enamel from extracted teeth is considered to be a useful biodosimetric approach for retrospective dose reconstruction (64, 65). Dose assessment using EPR technology in theory can provide a rapid assessment for radiological mass casualties, giving it an advantage over alternative dose assessment approaches (66–68). Using currently available technologies, the thresholds for the dose assessment EPR-based approaches advocated for triage screening (e.g., in vivo teeth, clipped nails) appear to span relevant life-threatening radiation doses (2–10 Gy).

Although assessment of in vivo EPR from teeth and ex vivo EPR from nail clippings from the extremities would not provide a complete mapping of a partial-body exposure, it would allow an estimate of the regional (head, extremities) radiation exposure and could point to bone marrow sparing. A mechanically induced EPR signal appears in nails after clipping that can overwhelm the radiation response, but soaking in water reduces the interference (69).

There are still technological challenges to be overcome before EPR will be ready for use in triage or clinical dosimetry. Multiple measurements would be necessary to limit the error of the prediction. The device for measurements teeth in vivo requires further development before it will be ready for fielding in a mass-casualty situation (67). The effects of sampling delay on the EPR signal in irradiated nails are not known, and unlike teeth, the signal in nails fades over a few weeks.

The use of optically stimulated luminescence (OSL) of tooth enamel has been suggested as an alternative to EPR⁹ (70, 71). While this approach eventually might be applicable to in vivo analyses, and current research suggests its applicability to doses in the triage range, it is still very early in its development [see refs. (71–73)].

Integration of Approaches

As described above, many approaches for partial-body biodosimetry are available. They each have advantages and disadvantages. Although each can be improved and developed further, a multi-faceted approach would optimize the assessment of radiation exposure (24, 46–48, 59). Clinical decisions will require multiple integrated end points. Using all the approaches (cytogenetics, biomarkers, gene expression profiles, clinical signs and symptoms, cutaneous injury bioindicators, and biophysical dosimetry) might overcome the gaps inherent in each technique. No single measurement will be adequate to provide a reliable diagnosis or to predict organ involvement. Moreover, this diagnostic approach should integrate two concepts. On the one hand, the concept of a multi-organ involvement in radiation-induced pathologies must be taken into account, even for radiation doses that are not life-threatening. On the other hand, the possible implications of abscopal effects need to be evaluated. It will be necessary to develop protocols that would make the combined approach an efficient one. Bertho et al. (59) showed effective integration of clinical signs and symptoms with biomarkers in the care of patients with serious partial-body radiation exposures. Ossetrova (46) demonstrated that very good discrimination of dose can be obtained by using multivariate analysis of three to five proteomic end points selected from distinctly different pathways. The combined use of proteomic and hematology biomarkers analyzed by multivariate analysis showed excellent discrimination using single doses in an NHP model (47, 48) and dose dependence using a mouse radiation model (74). Inclusions of clinical signs and symptoms with molecular biomarkers have merit. Given the tremendous heterogeneity of the U.S. population, where healthcare providers will encounter, among others, a large number of chronically ill patients, many on multiple medications, confounding variables will make biomarkers an extremely useful adjunct. With this approach, the variability among individuals and the possibilities for non-specificity for radiation exposure become less of a concern. While this

As integrated algorithms are applied to radiation accident victims, they need to be re-evaluated retrospectively (see below). Additional samples should be acquired and stored to permit subsequent analysis of alternative markers in an effort to improve the capability for prediction of outcomes.

Rapid, accurate dose assessment will be critical to the medical response to an incident involving radiation exposure, particularly after an improvised nuclear device when potentially tens of thousands of people could benefit from acute radiation syndrome mitigation and hundreds of thousands of people will be concerned (the worried well) that they may have received dangerous doses of radiation. Advances in preparedness will require the development of multiple new approaches and tools for the initial triage of victims to allocate and dispense countermeasures to those who will need them to prevent, mitigate and/or treat the harmful effects of ionizing radiation and particularly to enable the most appropriate allocation of scarce resources. These tools are also necessary for psychosocial considerations of active reassurance to mitigate mass panic and risk assessment to prepare patients psychologically for potential health consequences such as cancer and long-term diseases.

Confounding Factors

The human population is quite varied; people are genetically unique and have disparate health status, diets, pharmaceutical intake and environmental exposures as well as differences in gender and age. All of these factors can modulate an individual’s response to radiation and the results of biodosimetry assays (40, 75). In addition, combined exposures of radiation with chemical toxicants or traumatic injuries and the nature of radiation exposure itself (acute or protracted, dose rate and radiation quality) can further confound the results of dose assessment assays. While these confounders influence assessment of total-body radiation as well as partial-body exposures, their impact is worth consideration in this context.

Further confounding dose assessment in the event of a large-scale incident, particularly after a nuclear detonation, are the effects that the environment will have on patient symptoms and their expression. Forced relocation of large numbers of people coupled with expected failures in potable water and food distribution systems could lead to large numbers of patients presenting with gastrointestinal illnesses that could mask or be mistaken for the symptoms of gastrointestinal syndrome. Psychological and stress-induced illnesses can also manifest themselves in ways that could mask physical symptoms due to radiation exposure. In these cases, biomarkers will inform clinical decisions by providing additional data that can differentiate these hundreds or thousands of patients.

Post-event Medical Follow-up

Until they are tested in a real event, it will be difficult to know whether any individual biomarker or cluster of end points discussed in this report is a good predictor of outcomes, including late effects. The feasibility of using any approach in radiological mass-casualty situations will be known only after the effort has been made to use it during such a situation. Assessment tools and treatment approaches will benefit from a retrospective analysis of how well they worked and how to optimize them for subsequent events. The Radiation Injury Treatment Network (RITN) has developed detailed contingency plans for evaluation and management of radiation victims who have severe myelosuppression (see http://www.nmdp.org/RITN/ and http://bloodcell.transplant.hrsa.gov/ABOUT/RITN/index.html). Detailed event and clinical data collection has been incorporated into their efforts to learn from each experience and improve on plans and capabilities (76, 77).

Gaps in knowledge and plans will be more evident in retrospect. To improve the medical capability for victims of radiation injury, follow-up and communication within the medical and scientific communities will be essential. To achieve these goals, data collection systems will be needed to integrate information on accidents. A central mechanism will be required to keep track of the interventions that have been found to be effective and the biodosimetry tools that worked.

This requirement is not unique to partial-body exposures (78). Following the health of the affected population will be critical to understanding the efficacy of both biodosimetric approaches and interventions for the improvement of subsequent events.