The enigmatic mechanism of irradiation-induced damage to the major salivary glands


  • RM Nagler

    1. Department of Oral and Maxillofacial Surgery, Oral Biochemistry Laboratory and Salivary Clinic, Rambam Medical Center and Faculty of Medicine, Technion-Israel Institute of Technology, Haifa, Israel
    Search for more papers by this author

Dr RM Nagler, Oral and Maxillofacial Surgery Dept., Rambarh Medical Centre, Haifa 31096, PO Box 9602, Israel. Tel: 972 6 644 2003, Fax: 972 6 654 1295, E-mail:


Irradiation is a central treatment modality administered for head and neck malignancies. Its major and most devastating side-effect is an induced damage to the major salivary glands. This article aims at suggesting a comprehensive explanation for the underlying mechanism of this damage, which has been considered as enigmatic throughout the 90 years since it was first described in 1911. The mechanism suggested is based on the considerable literature concerning this enigma in rat salivary glands. According to this proposed mechanism, the irradiation results in a sublethal DNA damage, which manifests and becomes lethal at a delayed phase. Thus, when the acinar progenitor cells are going through a reproductive phase when parenchylmal replenishment is required, they die. The injurious agents, which result in this delayed reproductive cell death, appear to be highly redox-active transition metal ions, such as iron and copper. These metal ions, which seem to be associated with secretion granules, are not necessarily contained within the granules as previously suggested, but rather are probably located at sites more proximal to the DNA.


Irradiation is a major treatment modality administered for head and neck cancer. Unfortunately, this modality does not spare patients from devastating consequences, prominent among which is the induced damage to the salivary glands which results in salivary hypofunction and consequent xerostomia, a major cause of distress. As noted in an earlier article in this series, the severe negative impact that salivary hypofunction has on the patient's life results from various secondary effects. These include impairment of taste, mastication, swallowing, speech and sleep patterns. Furthermore, a reduction in saliva leads to diminished protection of the oral cavity against injuries to both hard and soft tissues, alters microbial flora to a more pathogenic type, precipitates a dry ulcerated painful mucosa and affects the wearing of oral prostheses.

The last decade was characterized by great improvement in the technology for delivering therapeutic radiation, mainly in using conformal and intensity-modulated irradiation (IMRT). This approach allows higher rates of salivary gland sparing in head and neck-irradiated patients (Eisbruch et al, 2000, 2001). However, and unfortunately for too many patients, this is not enough. Their salivary glands are inevitably irradiated while treatment for their cancer is administered. This occurs because of the location of the glands (symmetrical and extended), the extensiveness of the tumors and the high doses of irradiation that are often indicated. Furthermore, there is no adequate treatment for radiation-induced salivary hypofunction currently available. One may speculate (and hope) that a better understanding of the damage mechanism involved may facilitate developing a proper treatment or preventive strategy.

Irradiation-induced damage to the salivary glands was first described by Bergonie (1911). Although the phenomenon has been studied extensively (a literature search revealed that since 1966 more than 900 studies concerning the phenomenon were published), its underlying mechanism is still considered an enigma and its devastating outcome remains unresolved. The accepted description of this entity in humans points to the high radiosensitivity of the parotid gland, in contrast to the relative radioresistance of the other major salivary gland, the submandibular gland (Kashima, Kirkham and Andrews, 1965; Baum et al, 1985). Various studies have revealed an over 50% reduction in parotid gland function within a few days following low irradiation doses of 2.5–10 Gy to the head and neck region. Eventually, the hypofunction exceeds 90% and the residual secreted whole-saliva obtains mucous-like properties (Funegard et al, 1994; Cooper et al, 1995; Liem et al, 1996; Funegard et al, 1997; Guchelaar, Vermes and Meerwaldt, 1997; Fox, 1998; Rode et al, 1999; Roesink et al, 1999; Taylor and Miller, 1999). Numerous morphological studies have demonstrated the destruction of serous cells with sparing of mucous cells (Cherry and Glucksmann, 1959; Glucksmann and Cherry, 1962; Dreizen et al, 1976; Stern et al, 1976). The high radiosensitivity of serous cells accompanied by the loss of serous-like properties in the saliva is the basis for the explanation of the high radiosensitivity of the parotid glands; there is a much higher prevalence of serous cells in the parotid gland as compared with the submandibular gland.

Most of the data available regarding this phenomenon are derived from numerous studies conducted with the rat head and neck irradiation model. This model has been used frequently because of its convenience and general similarity to human salivary glands with respect to the general mechanism of fluid and protein secretion, and because of the rapid reduction in salivary flow following irradiation. However, there are clear anatomical and physiologic differences between salivary glands of humans and rats. For example, saliva is secreted spontaneously (i.e. basally) in humans, but only following stimulation in rats.

There are also important differences with respect to the response of salivary glands to irradiation. For example, it is well established that following irradiation in humans there is an immediate serous cells death that is profound and is accompanied by an inflammatory cell infiltration, which does not occur in rats. This is followed by a continuous reduction of salivary flow rates. In contrast, in the rat model a transient and partial recovery phase occurs at 2 weeks following irradiation and the continuous salivary flow rate reduction occurs only afterwards. Nonetheless, there is a huge body of data on irradiated rat salivary glands, and these data can provide key insights into the damage mechanism that gives rise to salivary hypofunction. The purpose of the current review is to partially resolve the enigma by suggesting a comprehensive underlying mechanism for this phenomenon, based on the numerous available studies.

Rat model studies – reports published in the last decade

In this section, the results from research studies, principally from two research groups, will be summarized.

In several papers published we have described the effects of 2.5–15 Gy head and neck X-irradiation on salivary secretion over a 1-year time frame (Nagler, Baum and Fox,1993a, b, 1996; Nagler et al, 1998a). Following irradiation with 15 Gy, parotid gland function (pilocarpine-stimulated secretion) is clearly reduced as early as days 3–14. This is followed initially by a nearly complete recovery to normal salivary flow rates. Thereafter, a delayed and profound additional reduction phase occurs. Thus, the first 2 weeks of the post-15 Gy irradiation phase in the rat parotid is characterized by indirect effects of irradiation, i.e. through a secondary and transient effect of mucositis leading to a reduction in eating, drinking and masticating. Interestingly, no such immediate response has been noted for the submandibular gland. By day 40 postirradiation, we observed a loss in salivary secretion for both the parotid (P < 0.01) and submandibular (P < 0.05) glands. Although a further gradual loss of function was demonstrated later in both glands, the parotid exhibited earlier and more severe damage. No immediate acinar cell loss was demonstrated for either of the glands at day 3. However, acinar cell loss was observed for the parotid, and less for the submandibular glands, by day 40. After 9 months, parotid secretion was reduced by 83% (P < 0.01), while that of the submandibular gland was reduced by 68% (P < 0.01). Rats irradiated with 15 Gy in the head and neck did not survive longer than 9 months in our studies.

On the third day following irradiation with 10 Gy, parotid function was reduced by 47% (P < 0.05), while no change in submandibular gland secretion was observed. By day 40, salivary function was at 44% of control levels for both the parotid (P < 0.01) and the submandibular (P < 0.05) glands. Thereafter, the secretion from both glands continued to decrease, with the parotid gland consistently demonstrating more severe damage. By 12 months postirradiation, both glands exhibited a profound functional loss, 76% (P < 0.01) for the parotid gland and 62% (P < 0.05) for the submandibular gland.

Following irradiation with 2.5 Gy we observed a much-delayed, highly significant functional reduction in both types of salivary glands. It is important to recall that head and neck-cancer patients are typically treated with a daily dose of 2–2.5 Gy. The diminished secretion of both the parotid and submandibular glands were not statistically significant up to 9 months. However, by 12 months post-2.5 Gy irradiation the secretory reductions in both the parotid and submandibular glands were of a substantial magnitude, 43 and 37% (P < 0.01), respectively. It is particularly noteworthy that this relatively small dose of irradiation resulted in 57% (P < 0.05) of the maximal parotid functional loss after 15 Gy and 59% (P < 0.05) of the maximal submandibular functional loss.

Konings, Vissink, Peter and colleagues in the Netherlands conducted a series of fundamental papers on irradiation effects on rat salivary glands (Peter et al, 1994a, b, 1995; Zeilstra et al, 1994; Coppes et al, 1997a, b 2000; Paardekooper et al, 1998; Roesink et al, 1999). They have succeeded in protecting rat salivary glands from irradiation damage over 6–30 days apparently by stimulating the proliferation of progenitor and/or stem cells and by increasing the compensatory growth potential of non-irradiated cells. This was achieved by administering various muscarinic and adrenergic agonists, such as isoproterenol, phenylephrine, methacholine, and pilocarpine. Of particular importance, these researchers concluded `the secretory granules do not play the often-assumed important role in the radiosensitivity of the salivary glands' (Peter et al, 1994b, 1995). Additional results from this group are consistent with our reported results that `the lack of cell loss soon after irradiation indicates that the reduction in gland function is caused by a compromised functioning of the acini' and that `the later loss of cells is probably due to death of cells that normally proliferate, leading to a further reduced secretory capacity' (Zeilstra et al, 2000). Other investigators have also reported a difference between short and long-term effects of irradiation on salivary glands in the rat model (Gustafsson et al, 1995; Funegard et al, 1997).

In summary, the work of the Dutch group suggested that the underlying mechanism of long-term damage irradiation to the salivary glands stemmed from an unmasking of latent lethal damage which occurred while replacement of dead cells by functionally intact cells took place. Moreover, the irradiation effect did not seem to be related to the secretory granules of the serous cells although these cells were more sensitive to irradiation.

Conclusions from these studies

Based on the above rat model studies, the following conclusions may be drawn:

(1) In the first 2 weeks following X-irradiation delivery to rat head and neck region there is a reduction in parotid gland function, starting almost immediately (∼4 h) and reaching a nadir toward the end of the first week (81% reduction on the eighth day following 15 Gy irradiation). An almost complete recovery toward normal function occurs by the end of the second week. During the 2 weeks following irradiation, no reduction in submandibular gland function is seen. Thus, in contrast to humans, no immediate direct effects of irradiation are detected in rats.

(2) At later time points, however, a delayed effect is observed and a reduction in secretory function noted for both gland types. This is seen up to 1 year following radiation, although most of the reduction is achieved within 3 months (Nagler et al, 1998b). A similar result has been reported in rhesus monkeys (Price et al, 1995).

(3) At any time points studied within 1 year, the irradiation effect on the parotid gland, i.e. salivary hypofunction, is greater than that on the submandibular gland. However, with time, the gap becomes smaller (Nagler et al, 1993a, 1996, 1998b; Hiramatsu et al, 1994).

(4) Extreme salivary radiosensitivity (permanent damage with as little as 2.5 Gy) is seen 1 year following irradiation. This dose caused a 43 and 37% hypofunction (P < 0.01) for rat parotid and submandibular glands, respectively, while a dose of 10 Gy caused a 76 and 62% hypofunction, respectively (Nagler et al, 1998b).

(5) No significant alterations in the composition of either the parotid or submandibular salivary secretions were found during the 9 months following irradiation, except for transient changes in the parotid gland, secondary to systemic malnutrition and dehydration effects, which are known to effect parotid gland (Nagler, Nagler and Laufer, 1997a).

(6) Morphometrical analysis at 40 days following 15 Gy irradiation showed that stromal cells replaced the parotid serous cells. In contrast at 3 days there was only marginal loss of serous cells, which did not correlate with the observed profound functional loss (Nagler et al, 1998a).


The short-term effects of irradiation on the parotid gland during the first 2 weeks is transient and secondary to the oropharyngeal syndrome. In rodents, this syndrome is characterized by severe transient mucositis resulting in dehydration, malnutrition and reduced mastication. These indirect changes are known to induce profound hypofunction of the parotid gland, and are unrelated to a direct effect on the secretory parenchyma (Quastler, Austin and Miller, 1956; Goepp and Fitch, 1962, 1963; Goepp, Fitch and Doull, 1967). Indeed, this suggestion has been experimentally tested in a study in which the 2-week irradiation effects were mimicked by employing a pair-feeding design with the animals (Hiramatsu et al, 1994; Nagler et al, 1996). Although some immediate cell death cannot be excluded, it does not seem to play a major role in eventual long-term damage, because of the nearly total recovery from the short-term effects in rats and the common lack of long-term recovery in humans (Nagler and Nagler, 1996). The immediate cell death seems to be much more severe and significant in primates than in rodents (Kashima et al, 1965; Stephens et al, 1986a, b, c), although the difference in radiosensitivity between the species is not yet understood. Stephens et al (1989, 1991) and `s-Gravenmade and Vissink (1992), have suggested this phenomenon may reflect an apoptotic cell death induced by irradiation. In an in vitro study of irradiation with the HSG salivary cell line, we were unable to see characteristics of apoptotic cell death (Nagler et al, 1995). This in vitro finding is supported by a more recent in vivo rat study (Paardekooper et al, 1998). It now appears that the most likely mechanism to account for immediate interphase cell death is that suggested earlier by (Vissink, Down and Konings, 1992), i.e. one predominantly caused by irradiation damage of membranes via lipid peroxidation.

Is there a common link that can account for the delayed effect on the one hand and the specific radiosensitivity of the parotid gland on the other? The existing paradigm for the radiobiological mechanism of delayed expressed damage is thought to involve latent DNA damage manifested during mitosis in cells with a low mitotic rate. This damage results in a reproductive cell death. The mitotic rate of rat salivary parenchymal cells is reported once every 1–3 months, with a cell division rate in the parotid twice as high as that of the submandibular (Cherry and Glucksmann, 1959; Glucksmann and Cherry, 1962). The accumulative nature of the delayed hypofunction of both glands, plus the fact that a major component of gland dysfunction is observed at 3 months postirradiation, as well as a asynchronous appearance of the submandibular effects, seems to be in accord with this data.

DNA is considered to be a very radiosensitive cellular target and has been shown to be so in the salivary glands as well (Sasaki and Toda, 1972; Sasaki and Nagami, 1973; Furuno, Iwasaki and Matsudaira, 1974). The `sudden' disappearance of normally functioning serous cells, as demonstrated by morphology, is consistent with an accumulative reduction in the volume of secreted saliva whose composition is preserved (Nagler et al, 1997a). Moreover, the unaltered expression pattern of salivary functional tissue-specific genes, such as amylase, proline-rich protein and kallikrein also is consistent with a reproductive cell death mechanism. The irradiation-induced injury to DNA, leading to such reproductive cell death, also seems reasonable in view of the profoundly high expression of DNA damage-induced (DDI) genes, such as c-Fos and jun-B (Mertz et al, 1992; Nagler and Nagler, 1996) following irradiation. Indeed (Vissink et al, 1992) conclude that `the later changes in salivary gland function are probably mainly dependent on repopulation of surviving stem cells'.

Two more puzzling questions must be addressed: (1) Why is the parotid gland preferentially affected? and (2) Are there any DNA-targeting agents, which increase the effect of irradiation on the salivary acinar cells? According to the hypothesis suggested by Abok et al (1984), heavy metal ions such as Zn, Mn and Fe contained within the secretion granules of the serous cells may be the damage-inducing agents that are responsible for the irradiation-induced immediate death. The parotid gland is composed primarily of serous cells, which contain high levels of such granules. In any case, two components of this hypothesis are puzzling in light of certain relevant facts:

(1) The predominant effect of irradiation on rat salivary glands is an induced delayed cell death and not an immediate one.

(2) According to basic principles of radiobiology, heavy metal ions as such cannot participate in the enhancement of irradiation-induced biological damage mediated via a profound production of hydroxyl free radicals. For a metal ion to be involved in such a process, three conditions must be fulfilled: it must be a transition metal ion, redox-active in physiologic conditions, and in a `free' state able to participate in the process; Fe and Cu can fulfil these conditions, but Zn and Mn do not.

A possible mechanism

The following mechanism for the irradiation-induced specific damage to salivary glands is suggested. The injurious agents resulting in delayed serous cell death and leading to specific parotid gland radiosensitivity are highly redox-active, transition metal ions, such as Fe and Cu, that are in some ways associated with secretion granules. These ions enhance the lethal effect that irradiation has on DNA, resulting in a delayed reproductive cell death. The association between these metal ions and the secretion granules may or may not be physical (topological), i.e. the metal ions may not be necessarily contained within the granules as Abok et al (1984) originally suggested. Rather it may be that both granules and ions are mobilized in/from the acinar cells as a result of a common or parallel step in a signal transduction pathway.

It seems more intuitive that reproductive cell death would occur when the redox active metals are not sequestered within the granules but, rather, is located in intimate proximity to the DNA. In any case both the irradiation-induced reproductive salivary cell death mechanism and the lack of a direct association between the granules and the long-term sites of damage are strongly supported in several published studies by Konings et al (Vissink et al, 1992; Peter et al, 1994b, 1995; Coppes et al, 1997). Further credence for this mechanism is found in a report published recently by O'Connell et al (1999) in which a direct correlation was shown between the decrease in postirradiation submandibular flow rate and a decrease in the acinar cell volume. Moreover, in a recent paper Zeilstra et al (2000), who studied the X-irradiation (15 Gy) effect on rat submandibular gland, concluded that `the later loss of cells was probably due to the death of cells that normally proliferate, leading to a further reduced secretory capacity'.

Thus, immediate effects of metal ion-mediated enhancement of irradiation damage in cells may occur but this does not seem likely to play a major role in the long term underlying mechanism. A positive correlation has been found between protection of parotid function at 2 months postirradiation and pre-irradiation degranulation and redox-active metal ion mobilization out of the gland (into the secreted saliva) in the rat (Nagler, 1998a; Nagler et al, 1998b). On the other hand, a negative correlation in the submandibular gland was demonstrated as well with no protection, no degranulation, no metal ion mobilization and no redox activity of the metal ions secreted within the saliva (Nagler, 1998a; Nagler et al, 1993b, 1997b, 1998b). Many others have also shown a clear difference between the response of the parotid and submandibular glands to irradiation (e.g. Peter et al, 1995). Further, dissociation may occur between short-term parotid hypofunction and a concomitant preservation of the parenchymal cells (Nagler, 1998b).

Key remaining questions

(1) What is the location, and specific pathologic role, of these `free' metal ions in the parotid gland? Is it reasonable to assume that they are normally associated with the secretion granules and yet must be close to the DNA in order to induce any damage?

(2) What is the physiological role of redox active metal ions in the parotid glands? This has never been described or discussed previously and may be of unique importance.

(3) If the replenishment capacity of the parotid gland is abolished because of serous cell killing, can one assume that the serous cells act as their own progenitor cells? The nature of salivary stem or progenitor cells is still unclear.

(4) What is the mechanism leading to DNA damage in the submandibular gland?

If the rat parotid gland can be assumed to represent a reasonable model for the human circumstance, then the obtained experimental results reviewed above can be very discouraging in a clinical sense. A 2.5 Gy dose, which has a severe, delayed effect in the rat, is the approximate dose given to humans daily within a routine radiotherapy course for head and neck malignancies. This dose of irradiation is given as part of a fractionation regime based on the assumption that such a low dose spares the exposed non-malignant healthy tissues, such as the salivary glands. It is imperative that a method for protecting the salivary glands from even low doses of irradiation has to be found.