Volumetric changes of the parotid gland during IMRT based on mid-treatment imaging: implications for parotid stem cell sparing strategies in head and neck cancer

Figures & data

Table 1. Patient characteristics, QoL questionnaire results and parotid gland volumes.

	All patients (n = 63)
Age, median (range)	62 (14–81)
Sex, n (%)
Female	42 (67%)
Male	21 (33%)
Dry mouth score, n (%)
1	15 (24%)
2	24 (38%)
3	12 (19%)
Stage, n (%)
I – II	18 (29%)
III	4 (6%)
IVA – IVC	40 (63%)
Unknown	1 (2%)
Sub-site, n (%)
Larynx	18 (29%)
Nasopharynx	5 (8%)
Oral	11 (17%)
Oropharynx	19 (30%)
Parotid Gland	5 (8%)
Other/Unknown	5 (8%)
Concomitant Chemotherapy, n (%)
Carboplatin	5 (8%)
Cisplatin	15 (24%)
Yes	22 (35%)
Unknown	21 (33%)
Contralateral PG volume, mean (range)	26.0 cm^3 (6.8–48.3 cm^3)
Ipsilateral PG volume, mean (range)	26.4 cm^3 (7.2–49.3 cm^3)

The contralateral parotid gland is defined as the one contralateral to the primary tumour and therefore receiving a lower dose of radiation.

The ipsilateral parotid gland is defined as the one ipsilateral to the primary tumour and therefore receiving a lower dose of radiation.

Figure 1. Linear regression showing change in PG volume from baseline to mid-treatment as a function of the planned mean dose to the contralateral parotid gland (a) and ipsilateral parotid gland (b). The corresponding regression lines are shown with accompanying 95% confidence intervals.

Figure 2. Scatter plot comparing the mean dose to contralateral parotid vs. mean dose to contralateral parotid SC PRV, based on planned dose from pre-treatment CT. The red circles represent patients who reported severe dry mouth (QoL score 3 or 4), and the black squares represent patients who did not (QoL score 1 or 2).

Figure 3. Logistic dose-response models estimating the risk of severe xerostomia as a function of mean dose to contralateral PG (a) and mean dose to the contralateral PGSC PRV (b), based on planned dose from pre-treatment CT. Patients were binned into four groups by mean dose: <10 Gy, 10–25 Gy, 25–35 Gy, and >35 Gy. Uncertainty bars represent 95% confidence intervals of xerostomia risk (vertical) and mean dose (horizontal) for each bin.

Table 2. Univariable logistic regression models of patient-reported severe xerostomia.

	OR/Gy (95% CI)	p Value
Contralateral PG (pre-treatment)	1.089 (1.034, 1.148)	p = .001
Contralateral PGSC PRV (pre-treatment)	1.071 (1.027, 1.118)	p = .002
Contralateral PG (mid-treatment)	1.134 (1.000, 1.285)	p = .050
Contralateral PGSC PRV (mid-treatment)	1.172 (1.035, 1.326)	p = .012
Contralateral PG (summed dose)	1.100 (1.017, 1.181)	p = .016
Contralateral PGSC PRV (summed dose)	1.106 (1.027, 1.190)	p = .008

Univariable logistic regression models of patient-reported severe xerostomia based on the dose estimated from the full treatment course from the pre-treatment CT contours (n = 63), the dose delivered by mid-treatment CT based on the mid-treatment contours (n = 34), or the full treatment course estimated from the summed dose based on both pre- and post-treatment CTs (n = 34).

Data availability statement

The datasets generated during and/or analysed during the current study can be made available from the corresponding author on reasonable request.