A Machine Learning Model Approach to Risk-Stratify Patients With Gastrointestinal Cancer for Hospitalization and Mortality Outcomes.

PURPOSE: Patients with gastrointestinal (GI) cancer frequently experience unplanned hospitalizations, but predictive tools to identify high-risk patients are lacking. We developed a machine learning model to identify high-risk patients. METHODS AND MATERIALS: In the study, 1341 consecutive patients undergoing GI (abdominal or pelvic) radiation treatment (RT) from March 2016 to July 2018 (derivation) and July 2018 to January 2019 (validation) were assessed for unplanned hospitalizations within 30 days of finishing RT. In the derivation cohort of 663 abdominal and 427 pelvic RT patients, a machine learning approach derived random forest, gradient boosted decision tree, and logistic regression models to predict 30-day unplanned hospitalizations. Model performance was assessed using area under the receiver operating characteristic curve (AUC) and prospectively validated in 161 abdominal and 90 pelvic RT patients using Mann-Whitney rank-sum test. Highest quintile of risk for hospitalization was defined as "high-risk" and the remainder "low-risk." Hospitalizations for high- versus low-risk patients were compared using Pearson's χ2 test and survival using Kaplan-Meier log-rank test. RESULTS: Overall, 13% and 11% of patients receiving abdominal and pelvic RT experienced 30-day unplanned hospitalization. In the derivation phase, gradient boosted decision tree cross-validation yielded AUC = 0.823 (abdominal patients) and random forest yielded AUC = 0.776 (pelvic patients). In the validation phase, these models yielded AUC = 0.749 and 0.764, respectively (P < .001 and P = .002). Validation models discriminated high- versus low-risk patients: in abdominal RT patients, frequency of hospitalization was 39% versus 9% in high- versus low-risk groups (P < .001) and 6-month survival was 67% versus 92% (P = .001). In pelvic RT patients, frequency of hospitalization was 33% versus 8% (P = .002) and survival was 86% versus 92% (P = .15) in high- versus low-risk patients. CONCLUSIONS: In patients with GI cancer undergoing RT as part of multimodality treatment, machine learning models for 30-day unplanned hospitalization discriminated high- versus low-risk patients. Future applications will test utility of models to prompt interventions to decrease hospitalizations and adverse outcomes.

Outcomes after salvage for HPV-positive recurrent oropharyngeal cancer treated with primary radiation.

PURPOSE: HPV-positive oropharyngeal squamous cell carcinoma (OPSCC) carries a favorable prognosis for patients, yet nearly 30% of patients will experience disease relapse. We sought to detail patterns of failure, associated salvage therapy, and outcomes for patients with recurrent HPV-positive OPSCC. METHODS AND MATERIALS: This is a single institution retrospective study of patients with recurrent HPV-positive OPSCC irradiated from 2002 to 2014. The primary study outcome was overall survival (OS, calculated using the Kaplan-Meier method). Secondary aims included patterns of first failure with descriptive details of salvage therapy. Solitary recurrences were defined as initial presentation of recurrence in a single site (primary, neck or oligometastatic), and multi-site was defined as local and regional and/or multiple sites of distant recurrence. Survival outcomes were compared using the log-rank test. RESULTS: The cohort consisted of 132 patients. The median follow-up was 59 months for surviving patients. Estimated 2-year and 5-year OS rates were 47% and 32%, respectively. Comparative 2-year and 5-year OS rates were 65% and 46% versus 19% and 9% for the solitary group and multi-site group, respectively (p < .001). CONCLUSIONS: Patients with recurrent HPV-positive OPSCC experience 5-year survival of approximately 32%. However, patients with a "solitary" recurrence including disease at the primary site, neck or oligometastatic site have more favorable long-term outcomes.

Partial omission of bleomycin for early-stage Hodgkin lymphoma patients treated with combined modality therapy: Does incomplete ABVD lead to inferior outcomes?

Classical Hodgkin lymphoma (HL) patients achieve excellent outcomes; therefore, treatment de-escalation strategies to spare toxicity have been prioritized. In a large randomized trial of early stage HL patients, omission of chemotherapeutic agents including bleomycin from the standard ABVD (doxorubicin, bleomycin, vinblastine, dacarbazine) regimen was not found to be non-inferior; however the effect of partial omission is unknown. We investigated the effect of bleomycin omission on outcome for 150 early stage HL patients. At eight years, freedom from relapse was 99% for both patients who received complete or incomplete bleomycin, which is reassuring for patients requiring bleomycin omission due to toxicity.

Decreased heart dose with deep inspiration breath hold for the treatment of gastric lymphoma with IMRT.

We hypothesized that deep inspiration breath-hold (DIBH) and computed-tomography image-guided radiotherapy (CT-IGRT) may be beneficial to decrease dose to organs at risk (OARs), when treating the stomach with radiotherapy for lymphoma. We compared dosimetric parameters of OARs from plans generated using free-breathing (FB) versus DIBH for 10 patients with non-Hodgkin lymphoma involving the stomach treated with involved site radiotherapy. All patients had 4DCT and DIBH scans. Planning was performed with intensity modulated radiotherapy (IMRT) to 30.6 Gy in 17 fractions. Differences in target volume and dosimetric parameters were assessed using a paired two-sided t-test. All heart and left ventricle parameters including mean dose, V30, V20, V10, and V5 were statistically significantly lower with DIBH. For IMRT-FB plans the average mean heart dose was 4.9 Gy compared to 2.6 Gy for the IMRT-DIBH group (p < 0.001). There was a statistically significant decrease in right kidney dose with DIBH. For lymphoma patients treated to the stomach with IMRT, DIBH provides superior OAR sparing compared to FB-based planning, most notably reducing dose to the heart and left ventricle. This strategy could be considered when treating other gastric malignancies.

Lingual Tonsillectomy Likely Does Not Improve Outcomes for Squamous Cell Carcinoma of the Head and Neck From an Unknown Primary Site.

OBJECTIVE: Patients with mucosal squamous cell carcinoma (SCC) of the head and neck almost always have a primary site in the base of tongue or tonsillar fossa. Lingual tonsillectomy has recently been advocated as part of the diagnostic evaluation as opposed to directed biopsies of the base of tongue and thought to possibly result in an increased likelihood or cure. The purpose of this project is to determine whether this is probable. MATERIALS AND METHODS: We reviewed the medical records of patients treated with primary radiotherapy (RT) between January 1983 and March 2013. The outcomes were compared following RT in consecutively treated patients with either T1-2 base of tongue or unknown primary (cancer of unknown primary) SCC with predominantly level 2 adenopathy. RESULTS: At 10 years, there were no clinically significant differences in the 2 groups, in local control, regional control, freedom from distant metastases, disease-specific, or cause-specific survival. Overall survival at 10 years was improved with T1-2 base of tongue cancers but not for those with T0 N3 disease. The reasons for this are unclear. CONCLUSIONS: Tongue base biopsy (or lingual tonsillectomy) likely increases the probability of identifying an unknown primary in the base of tongue, but it does not improve outcome following RT for patients with cancer of unknown primary SCC with predominantly level 2 adenopathy.

How Does MR Imaging Help Care for My Breast Cancer Patient? Perspective of a Radiation Oncologist.

Radiation therapy is used in many cases of both early and late breast cancer. The authors examine the role of MR imaging as it pertains to radiotherapy planning and treatment approaches for patients with breast cancer. MR imaging can assist the radiation oncologist in determining the best radiation approach and in creating treatment planning volumes. MR imaging may be useful in the setting of accelerated partial breast irradiation. Radiation oncologists should attend to MR breast images, when obtained, to ensure that these imaging findings are taken into consideration when developing a radiation therapy plan.

Local control in non-metastatic medulloblastoma.

BACKGROUND: A single-institution review of long-term outcomes and factors affecting local control (LC) following radiotherapy for non-metastatic medulloblastoma. MATERIAL AND METHODS: From 1963 to 2008, 50 children (median age, 7.3 years; range 1.2-18.5) with stage M0 medulloblastoma were treated with radiotherapy; half underwent a gross total resection (no visible residual tumor) or near-total resection (< 1.5 cm(3) of gross disease remaining after resection). Median craniospinal dose was 28.8 Gy (range 21.8-38.4 Gy). Median total dose to the posterior fossa was 54.3 Gy (range 42.4-64.8 Gy). Eighteen patients (36%) received chemotherapy as part of multimodality management, including 11 who received concurrent chemotherapy. RESULTS: Median follow-up was 15.7 years (range 0.3-44.4 years) for all patients and 26.6 years (range 7.3-44.4 years) for living patients. The 10-year overall survival, cancer-specific survival, and progression-free survival rates were 65%, 65%, and 69%. The 10-year LC rate was 84% and did not significantly change across eras. Four percent of patients experienced local progression five years after treatment. On univariate analysis, chemotherapy and overall duration of radiotherapy ≤ 45 days were associated with improved LC. Patients receiving chemotherapy had a 10-year 100% LC rate versus 76% in patients not receiving chemotherapy (p = 0.0454). When overall radiotherapy treatment lasted ≤ 45 days, patients experienced a superior 95% 10-year LC rate (vs. 73% in patients treated > 45 days; p = 0.0419). Three patients (6%) died from treatment complications, including radionecrosis/cerebellar degeneration, severe cerebral edema leading to herniation, and secondary malignancy. CONCLUSIONS: While we cannot draw definitive conclusions given the retrospective nature of our study, our long-term data suggest that reductions in craniospinal dose and boost target volume to reduce toxicity have not compromised disease control in the modern era. Our data also support analyses that implicate duration of radiotherapy, rather than interval between surgery and radiotherapy, as a factor in LC. Chemotherapy in multimodality management of medulloblastoma may have an underappreciated role in improving LC rates.

Late toxicity following craniospinal radiation for early-stage medulloblastoma.

BACKGROUND: The purpose of this study is to review late toxicity following craniospinal radiation for early-stage medulloblastoma. MATERIAL AND METHODS: Between 1963 and 2008, 53 children with stage M0 (n = 50) or M1 (n = 3) medulloblastoma were treated at our institution. The median age at diagnosis was 7.1 years (range 1.2-18.5). The median craniospinal irradiation (CSI) dose was 28.8 Gy (range 21.8-38.4). The median total dose, including boost, was 54 Gy (range 42.4-64.8 Gy). Since 1963, the CSI dose has been incrementally lowered and the high-risk boost volume reduced. Twenty-one patients (40%) received chemotherapy in their initial management, including 12 who received concurrent chemotherapy. Late sequelae were evaluated by analyzing medical records and conducting phone interviews with surviving patients and/or care-takers. Complications were graded using the NCI Common Terminology Criteria for Adverse Events, version 4.0. RESULTS: The median follow-up for all patients was 15.4 years (range 0.4-44.4) and for living patients it was 24 years (range 5.6-44.4). The overall survival, cause-specific survival, and progression-free survival rates at 10 years were 67%, 67%, and 71%, respectively. Sixteen patients (41% of patients who survived five years or more) developed grade 3 + toxicity; 15 of these 16 patients received a CSI dose > 23.4 Gy. The most common grade 3 + toxicities for long-term survivors are hearing impairment requiring intervention (20.5%) and cognitive impairment (18%) prohibiting independent living. Four patients developed secondary (non-skin) malignancies, including three meningiomas, one rhabdomyosarcoma, and one glioblastoma multiforme. Three patients (5.6%) died from treatment complications, including radionecrosis, severe cerebral edema, and fatal secondary malignancy. CONCLUSION: Ongoing institutional and cooperative group efforts to minimize radiation exposure are justified given the high rate of serious toxicity observed in our long-term survivors. Follow-up through long-term multidisciplinary clinics is important and warranted for all patients exposed to radiotherapy in childhood.