Purpose/Objective(s) Molecular imaging enables improved understanding of the anatomic regions involved with metastatic prostate cancer (PC) due to the improved sensitivity compared to conventional imaging. Consequently, there is increasing awareness that PC metastasizes relatively frequently to the para-aortic (PA) lymph node region, compelling some radiation oncologists to electively cover this area if it is determined to be at high risk or if nodes are already involved. However, the optimal contouring boundaries are unknown. The objective of this study is to use molecular imaging to develop guidelines for contouring the PA clinical target volume (CTV) in patients with PC. Materials/Methods We conducted a retrospective cohort study of patients with PC undergoing 18F-fluciclovine (FLU) or 18F-DCFPyL PSMA PET/CT at our institution. Images of patients with PET-positive PA lymph node involvement were imported into the treatment planning system and the avid nodes were contoured. The radial distance from the epicenter of each node was measured in all dimensions relative to the inferior vena cava and aorta, and superiorly relative to the left renal vein. The PA CTV was delineated per the 2021 NRG/RTOG post-operative endometrial/cervical cancer atlas, as principles of these malignancies are frequently used to delineate this region. Descriptive statistics were used to assess the performance of this gynecologic atlas and generate recommendations for contouring in PC patients. Results 246 men had molecular PET imaging at our institution from 10/2016 to 1/2022 (48% FLU, 52% PSMA). 31 men (13%) had evidence of PA nodal metastasis. In these patients, 16% had de novo, 52% recurrent, and 32% castration-resistant metastatic PC. The median number of positive PA lymph nodes per patient was 1 (IQR 1-3.5). The distance from central vasculature and left renal vein are shown in the Table. Only 66% of PA nodes were covered using the left renal vein as the superior border and 75% were covered using 1.5 cm above the left renal vein. Overall, the gynecologic atlas missed 25% of nodes (57% due to left sided and 30% due to superior misses), suggesting unique target delineation should be considered for PC patients. We recommend expanding the PC PA CTV in the left lateral and superior dimensions, with detailed recommendations to be included in our presentation. Conclusion We used molecular PET imaging to determine the anatomic patterns of PA metastasis to develop contouring guidelines for creating a PA CTV in patients with PC. Although the optimal patient selection and clinical benefits of PA RT remain uncertain, our results will aid in delineating the optimal target and are currently being validated in an independent dataset. Table. Distance from central vasculature and left renal vein (cm)

months (31.3-60.6 months), 22 biochemical failures have been observed for a 4-yr bNED survival of 84.4% (95%CI: 77.9-90.9%).On UVA, bNED-survival after sRT was significantly more likely for patients with VUA-only lesions (VUA-only vs others, HR=0.307, 95%CI: 0.120-0.784,p=0.014) and with smaller lesions (for every cc, HR: 1.071, 95%CI: 1.025-1.119,p=0.002).These associations remained significant (p< 0.01) on multivariate analysis as well.For patients with VUA-only disease or with lesions smaller than 0.5cc, 4-yr bNED survival rates were 90.7% (95%CI: 83.4-98.0%)and 90.6% (95%CI: 83.9-97.3%),respectively.The 46 patients with both favorable features had a 4-yr bNED rate of 94.6% (95%CI: 87.3-100%).Conclusion: These data support local restaging with DCE-MRI before sRT in the setting of a biochemical failure after RP.Patients with VUA-only and/or small volume lesions have an excellent outcome after dose-escalated sRT.

Diabetic retinopathy is a known complication of diabetes (db) mellitus. Between 40% and 45% of patients diagnosed with db have been reported to have microvascular retinal changes. Radiation-induced retinopathy (RIRN) is a known delayed complication in head and neck cancer (HNCa) patients receiving radiation therapy (RT) +/- chemotherapy. The goals of this study were as follows: (1) To compare the incidence of RIRN in a population of HNCa patients treated with RT to a subset of HNCa patients with preexisting db. (2) To examine the effect of pre-existing db in the incidence of RIRN. (3) To evaluate the effect of pre-existing db on the dose parameters (TD5, TD10, TD20) associated with RIRN using logistic function modeling. The medical records of 220 HNCa patients who received RT +/- chemotherapy were retrospectively reviewed to record the incidences of preexisting db and post-RT RIRN. Median follow-up time was 5.9 years (range, 0.5-30.6 years). Minimum doses received by one-quarter of the retina that exhibited prominent signs of RIRN were estimated by treatment plan evaluation and used for analysis. A Fisher's exact test was used to determine the association between preexisting diabetes and RIRN. A logistic function was used to describe the dose response for the incidence of RIRN for patients treated with RT +/- chemotherapy. The model was fitted first to all patients and then to the subset of patients who did not have preexisting db. The maximum likelihood method was used to optimize the fit. The incidence of RIRN in all patients was 15.9%. The incidence of RIRN in the patients with pre-existing diabetes was 47.8%. Fisher's exact test indicated that preexisting diabetes was significant in the incidence of RIRN (p = 0.001). The estimated TD5, TD10, and TD20 for all patients (including those with pre-existing db) were 38.2 Gy, 49.6 Gy, and 62.0 Gy, and for the group excluding patients with preexisting db they were 49.8 Gy, 59.5 Gy, and 70.1 Gy. The estimated TD50 and γ50 for all patients were 83.2 Gy and 1.36, and for the subset of patients without preexisting db they were 88.2 Gy and 1.69. (1) The incidence of RIRN is statistically significantly higher (p = 0.001) in patients with preexisting db compared to that in all patients. (2) Preexisting db affects the incidence of RIRN and dose-response parameters. (3) The decrease in TD5, TD10, and TD20 for RIRN in the group of all patients (including those with pre-existing db) when compared to the group excluding patients with preexisting db suggests that the preexisting db may decrease the dose threshold for the onset of post-RT retinopathy. Our analysis suggests that the subset of patients with preexisting db may be a high-risk group for post-RT retinopathy.

Purpose: To estimate and compare the doses received by the obturator, external and internal iliac lymph nodes and point Methods: CT-MR fused image sets of 15 patients obtained for each of 5 fractions of HDR brachytherapy using tandem and ring applicator, were used to generate treatment plans optimized to deliver a prescription dose to HRCTV-D90 and to minimize the doses to organs at risk (OARs). For each set of image, target volume (GTV, HRCTV) OARs (Bladder, Rectum, Sigmoid), and both left and right pelvic lymph nodes (obturator, external and internal iliac lymph nodes) were delineated. Dose-volume histograms (DVH) were generated for pelvic nodal groups (left and right obturator group, internal and external iliac chains) Per fraction DVH parameters used for dose comparison included dose to 100% volume (D100), and dose received by 2cc (D2cc), 1cc (D1cc) and 0.1 cc (D0.1cc) of nodal volume. Dose to point B was compared with each DVH parameter using 2 sided t-test. Pearson correlation were determined to examine relationship of point B dose with nodal DVH parameters. Results: FIGO clinical stage varied from 1B1 to IIIB. The median pretreatment tumor diameter measured on MRI was 4.5 cm (2.7– 6.4cm). The median dose to bilateral point B was 1.20 Gy ± 0.12 or 20% of the prescription dose. The correlation coefficients were all <0.60 for all nodal DVH parameters indicating low degree of correlation. Only 2 cc of obturator nodes was not significantly different from point B dose on t-test. Conclusion: Dose to point B does not adequately represent the dose to any specific pelvic nodal group. When using image guided 3D dose-volume optimized treatment nodal groups should be individually identified and delineated to obtain the doses received by pelvic nodes.

Patients with nasopharyngeal cancers (NPc) often present with otitis media with effusion. The goal of this study is to assess the relationship between the location and extension of the nasopharyngeal tumor, pre- and postradiation therapy (RT) otitis media with effusion (OME) and post-RT sensory-neural hearing loss (SNHL). The medical records of 124 NPc patients who received RT +/- chemotherapy were retrospectively reviewed to record the incidence of pre- and post-RT OME and post-RT SNHL. The median follow-up time was 5.9 years (range, 0.5-22.5 years). Tumor location and extension were determined based on available information from patient records, magnetic resonance imaging, and/or computed tomographic information. Tumors were classified as follows: class I, tumor localized in the fossa of Rosenmüller; class II, tumors extended from the fossa to the parapharyngeal space; class III, tumors occupying the parapharyngeal space. The incidence of a morbidity was recorded when a patient had documented diagnosis of OME at the time of RT consultation or post-RT follow up. The criteria for persistent clinically relevant SNHL was defined as a 10-dB increase in the hearing threshold at a high frequency (4 kHz) evaluated by pure tone audiometry in two consecutive evaluations performed 6 months after RT. A chi-square test was used to analyze the association between location/extension of the tumor and pre-RT OME, pre-RT OME, post-RT OME, post-RT OME, post-RT SNHL, location of the tumor, and post-RT SNHL. The incidence of pre-RT OME increased with tumor location/extension as follows: class I, 15%; class II, 40%; and class III, 60%; and was associated with tumor location/extension (p< 0.026). The incidence of post-RT OME was statistically associated with pre-RT OME (p = 0.036). The incidence of post-RT SNHL was statistically associated with post-RT OME (p = 0.001). The incidence of post-RT SNHL was associated with tumor location/extension (p = 0.038). Our results indicate that location/extension of the tumor in the nasopharyngeal space contributes to the development of pre-RT OME. Pre-RT OME may increase the risk for post-RT OME. Post-RT OME may increase the risk for post-RT SNHL. An association was also observed between tumor location/extension and the incidence of post-RT SNHL. These analyses indicate that the location/extension of the tumor may be a factor in the occurrence of post-RT SNHL in nasopharyngeal cancer patients and may contribute to the risk of post-RT SNHL along with other known risk factors (e.g., RT dose, adjuvant chemotherapy, age of the patient).

Purpose: To evaluate the efficacy of image guided LDR Syed implant with differential loading for the treatment of cervical cancer. Methods: CT image sets of 20 patients with Syed applicator were used to generate 2 sets of treatment plans(40 Plans). Target volume (PTV) were delineated using CT/MR fusion. In 1st set of plans, uniform needle length and uniform loading was used as conventional method. In the 2nd set, the locations, lengths of the needles as well as the number, location and source activity were optimized using iterative variation of these parameters in context of anatomical information to obtain a coverage of PTV with a reference dose rate of 50 cGy/hr of 85% while limiting the dose to bladder, urethra and rectum. For both sets of the plans Conformity index (COIN), Homogenity index (HI), Overdose volume index (OVI) and doses to 2 cc of bladder, urethra and rectum, sigmoid were estimated. Results: For 1st set of plans COIN, varied between 0.47 to 0.60, for the 2nd, it varied between 0.60 to 0.72. For 1st set of plans HI varied between 0.4 to 0.52, for the 2nd, it varied between 0.3 to 0.0.4.For 1st set of plans OVI varied between 0.33 to 0.45, for the 2nd, it varied between 0.20 to 0.30.A 15–45 % reduction of dose to 2 cc of bladder and urethra, 20–40% reduction in dose to 2 cc of rectum and sigmoid were obtained in 2nd set of plans. Conclusion: Image guided brachytherapy with optimization of needle location, length and activity of Ir‐192 ribbons yields higher conformity of prescribed dose to target volume while reducing the doses to OARs.

To evaluate and compare the early toxicities after intraoperative radiation therapy (IORT) to those of BCHDRB brachytherapy for the treatment of breast cancer. A retrospective review of a prospectively collected dataset of early-stage breast cancer patients treated between 2007 and 2012 was performed to evaluate the early toxicities related to IORT as compared to BCHDRB. Seventy-eight patients were treated with IORT and 15 patients were treated with BCHDRB. The prescribed dose for IORT patients was 20 Gy to the surface of the applicator with the radius of the applicator varying from 3.5 to 5.0 cm. The dose at 1 cm from the applicator varied from 5 to 7 Gy. BCHDRB patients received a total dose of 34 Gy delivered with twice-daily fractionation over 5 days. The cavity size varied between 30 to 50 cc. The dose was prescribed to 1.0 cm from the surface of the cavity. The median follow-up time for the IORT group was 12.5 months and that for the BCHDRB group was 36 months. Toxicities were evaluated using the Radiation Therapy Oncology Group (RTOG) toxicity scoring system based on the Common Terminology Criteria for Advanced Events version 3.0; posttreatment cosmesis was assessed using the Harvard breast cosmesis score. Early postoperative seromas (grades I and II) developed in 26 (32.5%) patients in the IORT group with a majority of the seromas being asymptomatic and resolving in 4 to 6 months. Four (5%) of the patients required a single aspiration for symptomatic seroma. Of the BCHDRB group 5 (33%) patients developed seromas (grades I and II). Symptomatic seroma was observed in 2 (13.3%) patients and 1 patient required a single aspiration. Other observed toxicities in the IORT and BCHDRB groups, respectively, were as follows: dermatitis, 4 patients (5.1%) and 2 patients (13.3%); fibrosis, 1 patient (1%) and 3 patients (20.0%); and breast infection, 4 patients (5.0%) and 2 patients (13.3%). In the IORT group, breast cellulitis was managed by antibiotics in 3 (4%) patients. In the BCHDRB group, fat necrosis occurred in 2 (13.3%) patients. In the IORT group, at the12-month follow-up, 92% of patients had excellent or good cosmesis and 8% of patients had fair or poor cosmesis. In the BCHDRB group, at 26 months of follow-up 83% of patients had excellent or good cosmesis and 17% had fair or poor cosmesis. IORT has been shown to be an equivalent treatment in early-stage breast cancer as whole-breast irradiation. The benefits of IORT include a shorter treatment time, less dermatitis, fewer infections, and improved cosmesis over BCHDRB.

Post-radiation therapy (RT) chronic otitis media with effusion (OMEc) has been statistically associated with the incidence of sensory-neural hearing loss (SNHL) in head and neck cancer (HNCa) patients. The purpose of this study was to estimate and compare the parameters of clinical significance associated with the risk of OMEc and SNHL after fractionated RT for HNCa patients through logistic modeling. Radiation oncology and otolaryngology records of 395 HNCa patients who received RT were retrospectively reviewed to assess the incidence of OMEc and SNHL, using air and bone conduction thresholds for high frequency (HF) hearing at 4 kHz. Median follow up was 5.4 years (0.5-30 years). Mean doses received by the middle ear and cochlea were estimated by evaluating the treatment plans. A logistic function was used to describe the dose response for patients treated with once- and twice-daily fractionation (fx), patients treated with RT alone, and patients treated with chemotherapy (chemo)-RT for the incidence of OMEc and SNHL. The maximum likelihood method was used to obtain estimates of model parameters: D50 = dose at which 50% of patients developed a complication and γ50 = normalized slope at D50. Doses corresponding to 5% and 10% incidences of complications, TD5 and TD10, were calculated. For OMEc, the estimated TD5 and TD10 in the RT-alone patients were as follows: once-daily fx, 38.5 and 44.9; twice-daily fx, 38.3 and 46.2 Gy. For the chemo-RT group they were as follows: once-daily fx, 33.9 and 40.6; twice-daily fx, 36.6 and 44.2 Gy. The estimated TD50 and γ50 in the RT-alone group were as follows: once-daily fx, 63.5 Gy and 1.87; twice-daily fx, 69.4 Gy and 1.64. For the chemo-RT groups they were as follows: once-daily fx, 60.3 Gy and 1.68; twice-daily fx, 66.2 Gy and 1.65. For HF-SNHL, the estimated TD5 and TD10 were as follows: once-daily fx, 33.5 and 41.5; twice-daily fx, 38 and 44 Gy. TD5 and TD10 for RT-alone were 37 and 43; for chemo+RT they were 33 and 39 Gy. The estimated TD50 and γ50 were as follows: once-daily fx, 61.5 Gy and 1.55; twice-daily fx, 65.5 Gy and 1.56; TD5 and TD10 for RT alone and RT-chemo were as follows, 64 Gy and 1.54; 60.0 Gy and 1.46. The results show that TD5, TD10, and TD50 for OMEc and SNHL lie within a short dose range of each other, indicating the similarity in sensitivity of the middle ear and cochlea. Since the incidence of OMEc has been associated with the incidence of SNHL, and OMEc has been suggested to enhance the probability of SNHL after RT, TD5 and TD10 obtained by modeling clinical data offer dose limits to middle ear that may help decrease the incidence of OMEc and, probably, SNHL.

PURPOSE:To compare the dose to the high-risk clinical target volume (HRCTV)-D90 and dose-limiting structures with the dose prescription to point A (Manchester system) using conventional methods to image-guided treatment planning with dose optimization and a dose prescription to D90 of the HRCTV. METHODS:CT-MR-fused image sets for 16 patients (stage IA2-IIIB) with uterine cervix cancer were obtained for each of 5 fractions of HDR brachytherapy using tandem and ring applicators to generate 2 sets of plans. In the first set, the dose was prescribed to point A with a standard dwell weight pattern reflecting the Manchester system loading pattern. In the second set, the dose was prescribed to D90 of the HRCTV with graphical plan optimization. Parameters of clinical significance (HRCTV-D90 and dose received by 1.0, 2.0, and 5.0cc of the bladder, rectum, and sigmoid) were compared, as recommended by GEC-ESTRO. RESULTS:HRCTV-V for 6 patients was =30cc (largest radius, <2.0cm) and >30cc (largest radius, =2cm) for the others. For the first set of plans, when HRCTV-V was <30cc, the ratio of doses HRCTV-D90/dose to point A varied from 1.10 to 1.40. When HRCTV-V was >30cc, it varied from 0.75 to 0.95. In the second set of plans, HRCTV-D90 always received the prescription dose. Doses to 2cc of one or more organs at risk exceeded >10% of the limiting doses for 9 patients in the first set of plans but remained <10% in all patients in the second set of plans. A 10-25% dose reduction to 2cc of the bladder and 15-35% to 2cc of the rectum and sigmoid occurred in the second set of plans. CONCLUSIONS:Image-guided brachytherapy with dose optimization yields a prescribed dose to the HRCTV-D90 while maintaining doses to organs at risk close to their limiting value, offering a dosimetric benefit over the traditional method.

Purpose: To compare four NTCP models to predict dose‐response for the incidence of radiation‐induced optic neuropathy (RION) and retinopathy (RIRP). Method and Materials: Logistic, log‐logistic, Poisson‐based and probit models were used to describe dose‐response. Fits to four data sets were obtained: RION in 101 patients treated twice‐daily (BID) and 172 patients treated once daily (QD), RIRP in 78 BID and 108 QD patients. Doses used in dose‐response analysis were converted to isoeffective dose in 2‐Gy fractions using α/β=1.76 Gy for RION and 2.65 Gy for RIRP. Maximum‐likelihood profile method was used to obtain model parameter values, D50 and γ, and corresponding confidence intervals. Results: Within the dose range bounded by available clinical data, the model predictions were similar. For data sets spanning a broad range of incidence a reasonable consistency between model parameters was observed, however, the log‐logistic model consistently showed larger D50 and shallower normalized slope γ. Specifically, for incidence of RION among patients treated QD, D50 ranged from 94.2 to 104.7 Gy and γ from 0.88 to 1.41. Model parameters for RIRP in the QD group were D50 from 72.2 to 74.0 Gy and γ from 1.51 to 2.16; in the QD group from 72.2 to 74.0 Gy and from 0.84 to 1.20, respectively. A large variation in model parameters was observed for RION in patients treated BID who showed incidence of complications spanning the range up to 20%. D50 ranged from 96.3 to 125.2 Gy and y from 0.80 to 1.56. Conclusion: Log‐logistic model tends to lead to larger D50 and lower γ compared to other models. Statements regarding normal tissue radiosensitivity and steepness of dose‐response based on model parameters should be made carefully as the latter are not only model‐dependent but also sensitive to the range of complication incidence exhibited by clinical data.

To estimate parameters of clinical significance associated with the risk of severe dry-eye syndrome (DES) after fractionated radiotherapy (RT) for head and neck cancers by using NTCP modeling. We analyzed 2 datasets: our 2010 data (UF-2010) and data published by Parsons et al1. The UF-2010 retrospective dataset included 77 patients treated for primary extracranial head and neck tumors between 1965 and 2000 whose lacrimal apparatus/entire globe was exposed to fractionated RT with both once-daily (QD) and twice-daily (BID) fractionation. Accurate estimation of doses received by the cornea and accessory lacrimal glands is difficult and unreliable due to their shallow depth. The radiation dose received by the major lacrimal gland (LG) was used for analysis and estimated from patient-specific treatment plans using CT scans or CT-MRI fusion, when available, or, for patients treated earlier in the series, scaled CT-scans of Rando Phantom. The endpoint of the study was ophthalmologic diagnosis of severe DES leading to vision compromise (RTOG Acute Radiation Morbidity Scoring Criteria Grade 3 and 4 toxicities or NCI CTCAE Grade 2 and 3 toxicities). The logistic function used to describe the dose response for both datasets was p= 1/[1+ exp(-4γ50((D/D50)-1))], where γ50 is the normalized slope and D50 is the dose at which 50% of patients develop complications. The parameter values were estimated using the maximum-likelihood fitting method with asymmetric confidence intervals (CI) determined using the profile-likelihood analysis (95% CI). The average dose to the LG varied from 20 to 75 Gy for UF-2010 and 30 to 83 Gy for Parsons et al. For the UF-2010 QD data, with TD50 of 48 Gy and γ50 of 2.67, the estimated values of TD5 and TD10 for DES were 34.5 Gy and 38.1 Gy. With TD50 of 48.8 Gy and γ50 of 2.81 for the QD+BID data, they were 36 Gy and 39.3 Gy. For the Parsons et al dataset (QD+BID), with TD50 of 40.3 Gy and γ50 of 2.44, TD5 and TD10 were 28.9 Gy and 32.2 Gy. The estimated values NTCP45, NTCP50, NTCP55, and NTCP60 were as follows: for the UF-2010 QD data, they were 33.9%, 60.9%, 82.6%, and 93.5%; for the QD+BID data, they were 29.4%, 56.9%, 80.7%, and 93%. For the Parsons et al dataset, they were 75.7%, 93.1%, 97.2%, and 99.2%. The NTCP modeling for DES shows that the incidence of DES increases with an increase in the total dose to the LG. Although a threshold cannot be determined from these datasets, from the NTCP modeling of 2 largest datasets for DES, limiting the values of TD5 for DES varies between 28.9 to 36 Gy and for TD10 between 32 to 39 Gy. The risk of DES increases significantly for doses over 40 Gy to the LG.

Purpose: To evaluate the the effect of total radiation dose received by cochlea on the risk of radiation induced sensory‐neural hearing loss (RI‐SNHL) using the Lyman NTCP model. Methods and Materials: A retrospective analysis for RI‐SNHL (audiology, otolaryngology, and radiation oncology records) of 410 patients (820 total ears) treated for selected treatment sites of head and neck cancers were performed. Due to the small volume of the organ, the dose volume analysis was not performed, instead the response was evaluated as a function of the average dose to cochlea. RI‐SNHL was defined as 10 dB loss at high freq (4 kHz) or low freq (average of frequencies between 0.5–2.0 kHz). The Lyman NTCP model was used to describe the dose response. Parameters of clinical significance were estimated using the maximum‐likelihood method with confidence intervals determined from the asymptotic covariance matrix (95% CI). Results: The average doses to cochlea varied between 0.5 – 81 Gy. The estimated values of TD50, m, and TD5 (calculated) for high frequency hearing loss were 63.1±1.9 Gy, 0.263±0.043, and 36.0 Gy and those for low frequency loss were 81.3±5.5, 0.168±0.041, and 59.0 Gy, respectively. The calculated γ50 for high and low frequency SNHL were 1.52 and 2.38, respectively. Conclusion: Lyman NTCP model fit to the RI‐SNHL data for high and low frequency hearing loss showed significant variation in pattern of incidence of hearing loss with increase in the dose to cochlea. The probality of incidence of RI‐SNHL at high frequency starts at lower dose (TD5 = 36 Gy) and increases relatively slowly (γ50 = 1.52) compared with that for low frequency RI‐SNHL (TD5 = 59 Gy, γ50 = 2.38). The NTCP model fit to the clinical data suggests that the dose to cochlea be limited to 36 Gy to reduce the possibility of any RI‐SNHL complications.

To evaluate the effects of chemoradiation and fractionation on the risk of radiation-induced sensorineural hearing loss (SNHL) as a function of total dose to the cochlea, and to estimate parameters that are clinically significant for limiting the dose to the cochlea using Lyman NTCP model. Postradiotherapy (RT) SNHL in 410 patients (820 ears) treated for head and neck cancers was retrospectively analyzed. Due to the small volume of the cochlea, the response was evaluated as a function of the average dose to the cochlea. Average doses received by the ipsilateral and contralateral cochleae were estimated using patient-specific 3-dimensional (3D) treatment plans. For patients who did not have a 3D treatment plan, doses were estimated by generating patient-specific treatment plans using a proxy phantom. SNHL was defined as a 10-dB loss at high frequencies (4 kHz) or low frequencies (averaging 0.5, 1.0, and 2.0 kHz). Pre-RT audiometry records were used as the baseline when available; otherwise age-specific standards were used. Univariate and multivariate analyses were performed on a subset of patients. Lyman's NTCP model was used to describe the dose response. Parameters of clinical significance were estimated using the maximum-likelihood method. The average dose to the cochlea ranged from 0.5 to 81 Gy. Chemoradiation was significant on univariate and multivariate analysis (p = 0.03 and 0.006, respectively), but fractionation was not (p = 0.7 and 0.4). In the NTCP modeling the estimated values of TD5, TD10, TD50 (Gy), and γ50 for high-frequency SNHL were, respectively as follows: RT alone: 37, 43, 64, 1.54; chemoradiation: 33, 39, 60, 1.46; once-daily fractionation (QD): 35.5, 41.5, 61.5, 1.55; and twice-daily fractionation (BID): 38, 44, 65.5, 1.56. For low-frequency hearing loss the estimates were: RT alone: 60.5, 65, 81.5, 2.56; chemoradiation: 55, 60.5, 79.5, 2.21; QD: 59, 64, 80.5, 2.47; and BID: 58.5, 63.5, 82, 2.28. The average dose to the cochlea ranged from 0.5 to 81 Gy. Chemoradiation was significant on univariate and multivariate analysis (p = 0.03 and 0.006, respectively), but fractionation was not (p = 0.7 and 0.4). In the NTCP modeling the estimated values of TD5, TD10, TD50 (Gy), and γ50 for high-frequency SNHL were, respectively, as follows: RT alone: 37, 43, 64, 1.54; chemoradiation: 33, 39, 60, 1.46; once-daily fractionation (QD): 35.5, 41.5, 61.5, 1.55; and twice-daily fractionation (BID): 38, 44, 65.5, 1.56. For low-frequency hearing loss the estimates were: RT alone: 60.5, 65, 81.5, 2.56; chemoradiation: 55, 60.5, 79.5, 2.21; QD: 59, 64, 80.5, 2.47; and BID: 58.5, 63.5, 82, 2.28.