Introduction

Radiation-induced toxicities (RIT) threaten treatment tolerance, cause tissue injury, compromise quality of life, and increase healthcare resource use [1]. No group is more at risk of RIT than head and neck cancer (HNC) patients. Even with improved radiotherapy (RT) technology and techniques, this cohort consistently suffers both acute and chronic RITs [2]. None is more impactful than OM.

Many past attempts to identify an effective pharmacologic solution for radiation-induced oral mucositis (ROM) have failed. It seems that underestimating the biological complexities of ROM’s pathogenesis contributed to missteps in accurately identifying druggable targets. Gone are the days when ROM was assumed to be solely due to direct DNA damage in epithelial stem cells, that a single group of cells (endothelium) influenced another (epithelium) [3], that normal and tumor cells responded identically to radiation, or that the molecular path leading to tissue damage was accurately represented by a linear cascade.

As the complexity of ROM pathogenesis has been further defined, we have mapped the steps leading to and sustaining injury, developed a biological event hierarchy, and sequentially integrated biological signaling into our picture of ROM’s clinical course. With this understanding, better strategies to identify effective druggable targets have ensued.

But executing on those strategies has been difficult. The biological cascade that follows a single acute dose of radiation is complicated. PubMed contains over 7000 papers describing normal tissue response to RT and about 10,000 with keywords including “radiation” and “normal tissue”. Overwhelmingly, each manuscript focuses on a single biological molecule, cell type, organelle, or pathway, at one time point. A global picture is absent. To put this into perspective, imagine you were tasked with drawing a country map by piecing together detailed maps of individual cities. Now imagine that the city maps changed daily; you get an idea of the difficulties in comprehensively defining ROM pathogenesis in the context of a fractionated radiotherapy course. A working paradigm describing ROM’s pathogenesis must address three observations. First, not only does radiation impact cancer cells and normal cells in different ways [4], but there is variability depending on normal cell type [5]. Second, while initial ROM clinical changes are attributable to immediate direct cell injury, its time and fraction-dependent progression and sustainment are dominated by indirect signaling, enzymatic shifts, and protein and non-protein intermediaries [6]. In aggregate, these indirect effects are the major contributors to ROM development. Finally, the extensive complex of biological events and mediators which initiate, catalyze, and sustain radiation-induced injury is dynamic. The degree with which it impacts ROM reflects the extent of homeostasis loss as a function of cumulative radiation dose. Thus, direct damage can be reasonably modeled as a radiation dose response, while the indirect damage pathway is subject to intrinsic genomic, epigenomic, metabolomic, and microbiomic [7] influences.

Biological characterization of a moving target

Contrasting with many models used to study radiation biology, clinical fractionated radiation regimens provide a different dosing dynamic and cell and tissue challenge. The biological “movie” following a single radiation exposure only partially reflects the repeated pounding suffered by normal mucosa during treatment in which the biological impact of radiation represents the sum of cumulative dosing. Further, the biological responses of previously irradiated cells and tissues will differ from those of radiation naïve ones, as is well recognized in the setting of clinical reirradiation. Whatever signals or pathways are induced one day are supplemented, overlapped, and modified on subsequent days of treatment (Fig. 1). Following radiation, lingering effects of treatment also project into delayed or late clinical events.

Fig. 1
figure 1

Differences in the clinical trajectory of oral mucositis between a bolus and fractionated stomatotoxic challenge: one blow from a sledgehammer vs. cumulative hits from a small ball peen hammer. The acuity of SOM development and resolution in patients receiving high-dose chemotherapy such as in conditioning regimens for stem cell transplants contrasts dramatically with the slower onset, duration, and time to resolution observed in patients receiving typical regimens of fractionated doses of radiation for the treatment of head and neck cancers. While there is similarity in the underlying pathogenic mechanisms, the repetitive challenge of multiple radiation doses produces a cumulative biological effect that results in injury, i.e., one blow from a sledgehammer vs. repeated hits with a small hammer

Full size image

Mucositis is dynamic morphologically

The type of oral mucosa impacts ROM risk as it typically effects the movable or thin mucosa but spares the more keratinized tissue [8]. Mean turnover time for tissues at risk ranges from 14 days (buccal mucosa) to 20 days (floor of the mouth) [8]. Consequently, if a single radiation exposure sufficiently destroyed basal stem cells on day 0 to completely stop cell renewal, the phenotypic effect—ulceration—would be expected between 2 and 3 weeks later, precisely what is observed in patients treated with bolus doses of stomatotoxic chemotherapy (Fig. 2) [9]. This contrasts with the extended onset and course of ROM where small incremental RT fractions are administered over time [10].

Fig. 2
figure 2

Fractionated radiated challenge influences biological interactions and phenotype in the development of SOM. The biological sequence which delineates the pathogenesis of mucositis has been described as a linear 5-phase process (5,6): initiation (I), signaling (S), amplification (A), ulceration (U), and healing (H). The concept was originally proposed based on data obtained following challenges with bolus chemotherapy in humans or a single dose of high energy radiation in animals (Panel 1). The sequence also characterizes the biological response and course associated with each daily fraction of radiation experienced by patients being treated for cancers of the head and neck. Critically, however, in cases of fractionated dosing, the net clinical effect is not only a consequence of the linear events initiated by each fraction but more importantly by the vertical crosstalk which recognizes that the biological consequences of each fraction not only affect tissue horizontally but also modifies or is modified by cell function primed by earlier doses (Panel 2)

Full size image

ROM’s clinical trajectory is well-known [10]. Most patients receiving 2 Gy daily fractions manifest clinical evidence of reduced renewal (epithelial atrophy) about 2–3 weeks after radiation start and peak ulceration by 4–5 weeks. In the context of mean turnover time, this would suggest that a cumulative radiation dose of about 20–25 Gy is necessary to overcome physiologic defense mechanisms and irreversibly injure enough basal stem cells to tip the balance from maintenance of homeostasis to generation of ulcerative ROM (Fig. 2).

Ulceration heralds the start of a biological tug-of-war in which cell-signaling pathways favoring wound healing are triggered but face an ongoing bombardment of biological drivers of ROM, including additional radiation fractions, reactive oxygen species (ROS), and NF-κB-driven (nuclear factor-kappa B) inflammation.

Biological concepts: ROM and dominos

ROM pathogenesis consists of parallel, sequential, and staggered molecular events occurring in a temporal dimension. Were ROM simply the consequence of a series of interdependent occurrences, it might be modeled by a row of dominos where each domino represented a component in the pathogenetic path. Hit the first domino and the next domino falls, then the next, and so on until ulceration occurs. However, such a linear representation ignores the complicated relationships which characterize ROM’s biological processes. A more realistic arrangement would account for the intricacies, redundancies, and interactions that occur, perhaps a pyramid arrangement in which the first domino hits the next 2, those hit the 4 behind them, etc. But that still would not accurately reflect biological dynamics which might require that 2 of the dominos in row 10 could not only fall backward, but also sideways, or that a loo** row of dominos could project from the major group and circle back to re-knock over recovered dominos. Even this model fails to recognize that the size and weight of each domino might vary from patient to patient (genomic variability), that the space between dominos is not the same (epigenetics), or even that the force applied to each is not equivalent (amplifying potential of the microbiome).

An alternative concept of ROM pathogenesis

A hypothetical argument suggests that ROM pathogenesis describes the cumulative effect of two distinct pathways which occur in a semi-staggered sequence. The first pathway, the immediate injury pathway, induces direct damage to basal stem cells, whereas injury induced by the second pathway, the indirect pathway, reflects the failure of normal host defense mechanisms to keep up with biological challenges imparted by accumulating radiation doses. The two occur in parallel, but their contribution to ROM pathogenesis is staggered and not equivalent. Within and between each pathway are interactions between cells, their components, signaling molecules, transcription factors and their products, and positive and negative feedback loops. While basal epithelial cells are the target end organ, inflammatory and non-inflammatory cells in the submucosa are active participants serving as message generators and carriers, both during the active period of radiation therapy and as conduits for late tissue changes [11]. And the tissue environment is dynamic: Mucosa irradiated on day 1 is not the same days or weeks later after being conditioned by chronic radiation exposure. Once gap junctions are disrupted, the microbiome also becomes a participant. This all occurs in the context of patients’ baseline physiological state which is likely altered by biological activity emanating from the tumor response to irradiation [12].

The concept of radiation injury being mediated by direct and indirect mechanisms has been described [13] (Hall S et al. as an example). What is novel is the idea that trajectory and severity of RIT are driven by an irrevocable tip** point in the balance between radiation-associated damage mechanisms and the capacity of physiological protective mechanisms to meet that challenge.

The immediate injury pathway causes basal epithelial cell death in the absence of intermediaries primarily through direct ionization of DNA chemical bonds and DNA cleavage by hydroxyl radical from ionization of water, both generating radiation-induced DNA double-strand breaks (DSB) [14]. While DSBs only have an identifiable role in 30% of radiation injuries [53]. Functionally, Nrf2 binds to and regulates the expression of acute radiation expression (ARE) genes which encode for both antioxidant proteins, including SODs, and a second tier of antioxidant enzymes including glutathione transferase and metallothioneins [54].

MnSOD signaling catalyzes Nrf2 upregulation [48]. In MnSOD-silenced hepatocytes, MnSOD acted as a signaling mediator for Nrf2-related survival genes and that expression of Nrf2 and its nuclear translocation could be mediated by MnSOD signaling [48].

In a reciprocal mechanism that might be interpreted as a cellular attempt to accelerate the control of potentially toxic levels of ROS, ARE-bound Nrf2 activates MnSOD [55] in a manner that is consistent with the importance of redox sensitivity in regulating Nrf2 and NF-κB in the genesis of ROM [56]. Furthermore, not only are reductions in Nrf2 associated with a parallel response in SODs, but HO-1, a Nrf2 target gene, is linked to Nrf2’s ability to attenuate NF-κB expression [57].

Of interest relative to the clinical trajectory of develo** ROM is the finding that Nrf2 activation is radiation dose dependent. Activation has been noted at radiation fractions of 2 Gy [58]. However, unlike the first line enzymatic response which is triggered by NF-κB and AP1 within minutes or hours after the first dose of irradiation, Nrf2 antioxidant induction and response are slower. Reportedly, there is a 5-day delay before the Nrf2 antioxidant activation is observed. Possibly, the Keap1-Nrf2 pathway serves as a lifeboat against threatening levels of oxidative stress that have thwarted effective control by first line enzymes.

It is not just the nucleus that is targeted by the immediate and indirect pathways

The impact of radiation on membrane lipids also contributes to the trajectory of cell injury, although the extent and timing are not well established [59]. A membrane stress apoptotic pathway has been described in which lipid peroxidation impacts both cell and mitochondrial membranes [60]. In the case of the cell membrane, the key messenger mediating apoptosis is ceramide. DSBs induced by high cumulative doses of radiation can directly activate ceramide synthase with the consequent generation of ceramide. Additionally, the effect of ROS, including O2●-, leads to lipid peroxidation, sphingomyelinase activation, and hydrolysis of membrane sphingomyelin to yield ceramide. In both cases, ceramide targets the RAC1/MAPK pathway which leads to the expression of MAPK8 and caspases 1, 3, and 6 and stimulation of the autocrine death receptor pathway. The significance of MAPK8 is noteworthy as it provides another interactional mechanistic feature given its implication in TNF-mediated apoptosis [61].

Mitochondrial membranes are also impacted as they respond to radiation-induced cytochrome c, activate the caspase cascade, release Ca++, and produce pro-apoptotic proteins [62]. The lipid peroxidation noted in mitochondria is associated with an increase in membrane permeability [62] which may signal the organelle’s demise.

P53—tip** the balance?

Radiation-induced DNA strand breaks initiate a p53 DNA-damage response [19]. Triggered by ATM activation, p53 accumulates in the nucleus to initiate cell cycle arrest and promote survival and regulation of oxidative stress and the intrinsic apoptosis pathway [63]. P53’s impact on the cell’s response to oxidative stress varies with the level of ROS. During the initial stages of radiation, when ROS is modest, p53 signaling favors antioxidation by increasing antioxidant enzyme transcription including MnSOD [64]. However, once the extent of oxidative stress outstrips the ability of cytoprotective mechanisms to cope with increasing ROS levels as a consequence of cumulative radiation, the role of p53 shifts from promoting cell survival to protecting the organism from irrevocably damaged DNA. JNK signaling activates pro-oxidant genes such as p53-upregulated modulator of apoptosis (PUMA). Lipid peroxidation is enhanced and previously supported MnSOD activation is instead impaired through mitochondrial disruption [20].

Clinical implications: how does the complexity of toxicity pathogenesis impact risk assessment, druggable targets, and precision medicine?

ROM occurs in two phases. Immediate injury mediated primarily by DSBs begins right after the first dose of radiation. Manifestations of cell injury such as ROM are, at this stage, probably limited and reversible up to a threshold dose, perhaps around 30 Gy. Mechanisms associated with immediate injury are only modestly influenced by, or responsive to, intrinsic factors such as patient genomics.

The indirect pathway is, for normal tissue, catastrophic, most clinically meaningful, and the result of cytoprotective mechanisms being overwhelmed by radiation-induced stress. It occurs at a threshold dose of radiation at which the cell and tissue capacity to neutralize or reverse the physiologic impact of recurrent radiation challenges is surpassed. The indirect pathway differs from the immediate pathway in three major ways: First, the radiation threshold dose at which it is initiated varies among patients as it is the subject to intrinsic controlling mechanisms including genomics, metabolomics, epigenomics, and microbiomics; second, it is indirect and enabled by intermediates including mediators of oxidative stress, cytokines, and enzymes; and third, it is influenced by signaling from bystander cells [65].

The consequences of these differences provide insight into clinical behaviors, opportunities for ROM risk prediction, and intervention targeting (Fig. 3). From a hierarchical standpoint, the ability to contain ROS and blunt the downstream effects of key transcription factors seems an obvious strategy for interfering with the indirect pathway. In the face of out-of-control ROS levels, solely blocking downstream events is likely to fail. So this leads to four approaches: (1) reducing oxidative stress by scavenging free radicals; (2) combination agents which are pleiotropic, effectively mitigating ROS levels, transcription factor activation, and cytokine production; (3) extrinsic supplementation in which therapeutically administered “lifeboat” compounds metabolize excessive ROS; and (4) pharmacologically mediated expression or activation of ROS-controlling enzymes.

Fig. 3
figure 3

Mucositis is the consequence of the cumulative biological effects of the radiation challenge exceeding the intrinsic capacity of physiological protective mechanisms. Increasing the threshold for a gain in biological protection results in consequent increase in the threshold for manifestations of mucosal injury. Interventional strategies that supplement protective mechanisms and pathways or those that stimulate intrinsic mechanisms to optimize their effectiveness have proven to be most effective.

Full size image

There is evidence that all four strategies have merit. A scavenger, amifostine, was a product of an early US Army’s Antiradiation Drug Development Program. Its clinical potential was realized when in 1999 amifostine was approved as a salivary gland protectant from RT. Although amifostine’s mechanism of action has been primarily attributed to free radical scavenging, it also appears to act in the p53 pathway [66]. Amifostine’s effectiveness as an ROM intervention is unclear with conflicting trial results. Palifermin (KGF1) showed moderate effectiveness in reducing severe ROM incidence [67,68]; however, its use in solid tumors has been largely proscribed because of its potential impact on KGF-1 receptor-bearing cancer cells. Palifermin is biologically pleiotropic: Aside from its stimulating epithelial proliferation, it interacts with a range of mechanisms associated with RIT including NF-κB and Nrf2 [69].

SODs play a major role in mitigating radiation-induced ROS-mediated signaling and damage [29,30,51,70,71]. In 1987, radioprotection was observed in mice treated with intravenously administered SOD [72]. Soon after, SOD1 (orgotein) was shown to be clinically effective in treating RITs associated with head and neck radiation [73] and late toxicities in patients receiving pelvic irradiation [74]. Gene therapy using SOD2 plasmid/liposomes also protected mice from radiation-induced esophagitis [75]. More recently, a SOD mimetic (avasopasem manganese) effectively reduced the incidence and duration of severe ROM in patients being treated for oral and oropharyngeal cancers [76]. Avasopasem is currently in a large phase 3 ROM trial (NCT03689712) and a smaller phase 2 trial for radiation esophagitis (NCT04225026).

The final approach aims to optimize internal ROS defenses by provoking Nrf2 activation [77,78,79,80]. While ROS intervention has demonstrable importance as an interventional target, other related targets early in the ROM biological cascade (innate immune modifiers and NF-κB) may also be impactful [81, 82].

Conclusion

RITs such as ROM have two distinct, but interactive pathways, immediate/direct and indirect. An effective management strategy must consider both optimal drug targets and dosing schedules. A treatment which maximally attenuates the indirect pathway and acts at a key “chokepoint” is likely to favorably impact the severity, incidence, and course of ROM without negatively impacting tumor response.

There is no FDA approved drug therapy to mitigate ROM in HNC patients. RITs, including ROM, remain common, create suffering for patients, threaten optimum cancer therapy, drain healthcare resources, and share pathogenesis components and sequences. Thus, a mechanistic understanding of ROM’s pathogenesis is critical to successful development of therapeutic interventions.