Salivary gland function in nasopharyngeal carcinoma before and late after intensity-modulated radiotherapy evaluated by dynamic diffusion-weighted MR imaging with gustatory stimulation

Xerostomia is an important complication results from radiation-induced salivary glands injury, which has a considerable impact on patients’ quality of life [1‐4]. Nowadays, there are still some disadvantages and limitations of the existed different methods of evaluating xerostomia in clinical practice [5‐9]. In addition, different methods of evaluation are inconsistent, resulting in a worse comparability and reproducibility. Therefore, an effective method, especially noninvasively, to evaluate salivary glands function after radiotherapy (RT) is demanded.

Diffusion-weighted magnetic resonance imaging (DW-MRI), a noninvasive imaging technique, allows the depiction of molecular diffusion caused by Brownian motion in biological tissues. The water molecule diffusion leads to signal attenuation, which can be quantified as the apparent diffusion coefficients (ADCs) [10‐12]. The rest and stimulated ADCs represent the basic and reserve function of salivary glands separately, so the ADCs before and after stimulation can be used to fully evaluate the functional status of salivary glands. Many studies [10, 13‐20] have suggested that DW-MRI based on the glands’ morphological and functional changes [14, 18, 20‐22] has great potential in investigating the mechanisms of the salivary glands injury induced by radiation. However, the salivary glands’ pattern of reaction to gustatory stimulation in DW-MRI and changes in ADCs as a result of RT have no consistent results in previous studies [10, 13‐15, 23‐25]. And for the evaluation of xerostomia induced by RT, DW-MRI has been used to assess the parotid glands [13, 15, 26] and submandibular glands [20, 27, 28] based on rest and stimulated ADCs, but the sublingual glands have not been assessed. Another concern is that the relationship between the ADCs and clinical methods used for evaluation of xerostomia, such as salivary flow rates (SFR) and xerostomia questionnaire (XQ) scores were also remains unclear and have not been proven consistent previously [15]. Moreover, the thresholds of ADCs and SFR suggesting xerostomia are still lacking.

In our prospective study of 30 nasopharyngeal carcinoma (NPC) patients treated with intensity-modulated radiotherapy (IMRT), we evaluated the feasibility of dynamic DW-MRI used transient gustatory stimulation with lemon juice in evaluation of salivary glands function before and late after IMRT by compared and correlated the changes in ADCs with clinical methods, being the SFR and the XQ scores, for evaluation of xerostomia. We hypothesized that the ADCs correlated with SFR and XQ scores, which suggested the ADCs can reflect the salivary glands injury induced by RT accurately.

Thirty NPC patients, 17 cases were low differentiation squamous cell carcinoma, 11 cases were undifferentiated non-keratinized carcinoma, and 2 cases were non-keratinized carcinoma. One, six, 14, nine cases were I, II, III, IV phase respectively.

As a result of stimulation with lemon juice, an initial steady increase to peak during the first DWI scan (30 s) and subsequent decrease in ADCs was seen in all glands before and after RT. The Tmax before RT was statistically significant among the three salivary glands (p = 0.008), the Tmax of parotid glands was the earliest. Tmax of sublingual glands after RT was significantly earlier than that before RT (p = 0.004). The rest and stimulated ADCs of three glands after RT were all significantly higher than those before RT (p = 0.000). For both before and after RT, the rest ADCs was higher in submandibular glands (1.09 ± 0.06, 1.45 ± 0.14) than in parotid glands (0.95 ± 0.08, 1.11 ± 0.16) and sublingual glands (0.94 ± 0.07, 1.25 ± 0.14) (p = 0.000) (Table 2).

The ADCs IRs in sublingual glands (28.19 ± 16.61, 29.15 ± 18.39) was higher than in parotid glands (10.51 ± 6.44, 10.17 ± 8.33) and submandibular glands (7.35 ± 6.11, 9.47 ± 9.79) before and after RT respectively (p = 0.000), and the ADCs IRs in parotid glands (10.51 ± 6.44, 10.17 ± 8.33) was higher than in submandibular glands (7.35 ± 6.11, 9.47 ± 9.89) before and after RT respectively (p = 0.000). The ADCs IRs before and after RT had no statistical significance of the three glands respectively (p>0.05). Both of the rest and the stimulated ADC CRs had no statistical significance in the three glands respectively(p>0.05) (Table 2).

The rest and stimulated SFR of parotid glands, submandibular mixed with sublingual glands and full three glands after RT were lower than those before RT (p = 0.000). After stimulation, the SFR IRs of parotid glands and the full three glands after RT (7.30 ± 4.99, 6.07 ± 12.07) were higher than those before RT (2.75 ± 2.34, 3.93 ± 7.68) (p = 0.000). The absolute value of the rest SFR CRs in parotid glands and the full three glands were higher (61.60 ± 32.13, 82.46 ± 8.82) than the stimulated SFR CRs (17.82 ± 61.32, 59.84 ± 26.22) (p = 0.000). The absolute value of the rest and the stimulated SFR CRs in parotid glands were lower (61.60 ± 32.13, 17.82 ± 61.32) than submandibular mixed with sublingual glands (87.43 ± 14.00, 82.48 ± 16.84) (p = 0.000) (Table 3).

The cutoff value of the rest ADCs in parotid glands, submandibular glands and the sublingual glands between xerostomia and non-xerostomia groups were 1.01, 1.20 and 1.08 (p = 0.000). The cutoff value of the stimulated ADCs in parotid glands, submandibular glands and sublingual glands between xerostomia and non-xerostomia groups were 1.10, 1.32 and 1.32 (p = 0.000). The cutoff value of the rest and stimulated SFR of full three glands between xerostomia and non-xerostomia groups were 0.19 g/min and 0.79 g/min (p = 0.000) (Table 4).

A positive correlation existed between the rest ADCs of submandibular glands and the rest SFR of submandibular mixed with sublingual glands and the full three glands before RT (r = 0.425, p = 0.019; r = 0.470, p = 0.009) (Fig. 1). A positive correlation also existed between the stimulated ADCs and the stimulated SFR in parotid glands before RT (r = 0.366, p = 0.047)(Fig. 2). A negative correlation also existed between the stimulated ADCs of sublingual glands and the stimulated SFR of full three glands after RT (r = 0.381, p = 0.038) (Fig. 2).

For the correlation between ADCs and XQ scores, a positive correlation existed between the rest ADCs of parotid glands after RT and XQ scores (r = 0.382, p = 0.037), the stimulated ADCs IRs of sublingual after RT and the XQ scores(r = 0.462,p = 0.010), the rest ADCs CRs of sublingual and the XQ scores(r = 0.397, p = 0.030)(Fig. 3).

Even though modern radiotherapy techniques, such as the IMRT, allow a dose distribution highly conformal to the target volumes, which spare the salivary glands function after RT is not always possible [10, 31, 32], thus, the reduction of salivary glands function is still a common side effect of RT [33]. Xerostomia is an important complication results from radiation-induced glands injury, which has great impact on quality of life [1‐4]. Reduced salivary glands function and declined quality of life related to oral health are typical both in the acute phase (and reversible after the end of therapy) and the chronic phase and sometimes can be lifelong [1, 34]. Messmer et al. found that xerostomia did not change significantly over the period of 5 years based on 121 patients followed for 41–90 months after RT, and both the difficulties with speaking and eating condition worsened [3], the mechanism remains unclear, and the histological changes in the gland tissue during radiation are not accessible in vivo [35, 36]. At the same time, because of the high variability in salivary glands function among patients, it is important and necessary to individually assess and monitor the function of salivary glands using in vivo quantitative tools efficiently [37].

In the major salivary glands, the sublingual gland is the smallest and the measuring and calculation of it is difficult to some extent, which is the reason why the sublingual gland was not included in previous studies. In this prospective study, a DWI protocol was applied to patients with NPC before and late after RT at rest and after stimulation with lemon juice in parotid glands, submandibular glands and sublingual glands respectively. These results were compared with SFR and XQ scores. Our main findings were that an initial steady increase to peak during the first DWI scan and subsequent decrease in ADCs was seen in all glands before and after RT as a result of stimulation with lemon juice, suggesting a persisting capacity of the acinar cells to produce saliva after stimulation. The ADCs IRs before and after RT had no statistical significance in all three glands and the maximum ADCs after stimulation of the three glands were higher than the rest ADCs before and after RT respectively also support this view. To date, only ascorbic acid tablet [10, 14, 24] and lemon juice [23, 25] were used to stimulate the salivary glands. Some researchers [25] have used lemon juice with six tablets of 100 mg as a stimulant and other researchers [13] performed the gustatory stimulation with two 500 mg tablets of ascorbic acid – reported an initial increase and subsequent fluctuation in the ADCs, those results are in accordance with our study. However, Thoeny [38] and Dirix [14] used a 500 mg tablet of ascorbic acid as the stimulant and advised the patients to let the tablet dissolve but not to chew it in their mouths, which resulted in an initial decrease and subsequent increase in the ADCs [24, 39]. According to the previous studies, the pattern of response to gustatory stimulation in salivary glands varied widely. This may contribute to the difference of the species and amount of stimulant. The gustatory stimulation induced by ascorbic acid continuous during MR imaging due to ascorbic acid tablet dissolves completely after a median 23 min (range, 20-27 min), however, gustatory stimulation with lemon juice is more short-lived [24], and higher number of tablets of ascorbic acid or two instantly bitten tablets of ascorbic acid are likely to stimulate simultaneously more receptors than a single slowly dissolving tablet of ascorbic acid does [13]. Those differences between the ascorbic acid and the lemon juice and the amount of stimulant may have been responsible for these contradictory results.

The rest ADCs was higher in submandibular glands than in parotid glands and sublingual glands for both before and after RT. This finding was in accordance with previous studies [10, 14, 24, 40]. In the rest state, about two thirds of all saliva is produced by the submandibular glands, while the parotid glands produce less than one third [39]. Thus, the expected higher proportional amount of water in the extracellular space of the submandibular glands can explain the higher ADCs at rest [14]. The rest SFR in parotid glands was lower than the submandibular mixed with sublingual glands before RT also support this view. While the difference in rest ADCs may also be explained by the different histologic composition of the glands: the submandibular glands were mixed serous with mucous gland, whereas the parotid glands were purely serous [39]. Furthermore, the other possible contributing factor to the lower ADCs of the parotid glands is that adipose tissue in the parotid glands is higher than the submandibular glands [39]. The DWI technique relies on a fat-suppression technique, which can result in an overall decreased signal intensity of fatty tissues for all b factors.

In our study, the stimulated ADCs IRs in parotid glands was higher than in submandibular glands before and after RT respectively, which in accordance with the fact that most of the saliva is produced by the parotid glands in the stimulated state. The result that the Tmax of parotid glands was earlier than the submandibular glands and the sublingual glands also supported it.

The rest and stimulated ADCs increased after RT as a result of radiotherapy damage. Most studies reported the same result that a rise in ADCs in patients with xerostomia [13‐15, 36], which are accordance with our observations. Though the time of the follow-up MRI examinations is variable among all these studies (from a few weeks to several months after RT), it was concordantly suggested that the ADCs increased after RT. Acinar cells show early and late microscopic alterations in glandular tissue response to radiation, there are particularly changes indicative of cell death in the early phase, the hypo-vascularization, formation of fibrous tissue in the late phase [2, 41‐43], which leads to an decrease in the cell density and a greater increase of free water in the extracellular space, the motion of free water more faster and the ADCs increased. The rest and stimulated SFR of parotid glands, submandibular mixed with sublingual glands and full three glands after RT were lower than those before RT also indicated that the salivary secretary function decreased after RT. In the stimulated state, the amount of saliva was more than the rest state. However, the absolute value of SFR CRs in parotid glands and the full three glands at rest were higher than that of stimulated state, which also indicated that the radiation induced the acinar cells damage. The rest and stimulated ADCs CRs in the three glands separately had no significance also support this view.

The reason why the ADCs IRs in sublingual glands was higher than in parotid glands and submandibular glands before and after RT separately was that in the stimulated state, the saliva induced by the stimulant accumulated at the bottom of the mouth, which closest to the sublingual glands, and the volume of sublingual glands was the smallest among the three major glands, those reasons resulted a greater stimulus intensity of the unit area in the sublingual glands.

The median follow-up time was 16 months, and it was known that the salivary glands function may recover up 5 years after RT [44]. The SFR IRs after stimulation of parotid glands and the full three glands after RT were significantly higher than that before RT, and the Tmax of the sublingual glands after stimulation was earlier than that before RT, suggesting the retention and even hyperactivity of the salivary function after radiation, and the ability to repair and reserve to some extent.

When regarding the cutoff value of the rest ADCs before and after RT predicting the xerostomia, to our knowledge, there is no report in previous studies. In our study, we calculated the cutoff value of changes in ADCs and SFR, which provides a quantitative basis for predicting the xerostomia after RT. The 0.19 g/min of the rest SFR between xerostomia and non-xerostomia groups was very close to clinical standard in diagnosing xerostomia [45], which also identified the accuracy of the swab method in saliometry. The area under the curve, sensitivity and specificity were high in all of cutoff value, indicating the feasibility of establishing prediction model based on large samples.

Finally, we analyzed the correlation between the ADCs and the SFR, the ADCs and the changes in ADCs and the XQ scores respectively. The rest ADCs in submandibular glands was correlated with the rest SFR of submandibular mixed with sublingual glands and the full three glands before RT, the stimulated ADCs was correlated with the stimulated SFR in parotid glands before RT, both of them suggested the rest ADCs can reflect the secretary function of the salivary glands. And the increased ADCs after RT corresponding to the decreased SFR also indicated that the ADCs can reflect the secretary function of the salivary glands.

Both of the rest ADCs of parotid glands and XQ scores, the ADCs IRs of sublingual glands and the XQ scores, the rest ADCs CRs and the XQ scores had a positive correlation, which provided some evidences for the ADCs in predicting the severity of the xerostomia after RT. But the other changes in ADCs and the XQ scores had no significance, considering the reason that the xerostomia caused by radiation injury to salivary glands is a complex of subjective and objective factors, which is associated with the decreased saliva, the changes in saliva composition and viscosity and the tolerance of the patients. The XQ scores have certain subjectivity, and the increase in ADCs after RT is influenced by many factors, resulting in the deficiency of correlation.

The present study has several limitations. First, the study population was relatively small, which only reflect the data of some people in a certain region. Second, this study only analyzed the salivary glands of NPC patients more than 1 year and within 3 years after RT, while the salivary glands function and xerostomia are dynamic process after RT, and IMRT may have a long term effect on the salivary glands function in NPC patients. Thus, we propose that a large sample in multicenter in patients with different cancers in head-and-neck after RT with multiple follow-up periods and a long follow-up time be conducted to establish a predictive model, which help to establish an objective standard of xerostomia induced by RT based on DW-MRI.

The ADCs increased after RT both in rest and stimulated state, the rest and stimulated ADCs with lemon juice before and after RT correlate with the SFR and the XQ scores, suggesting that DW-MRI with transient gustatory stimulation is potentially useful to evaluate hypofunction of salivary glands and also is a useful tool for detection of xerostomia induced by radiation noninvasively and objectively in clinical practice.