Traditional radiomics involves the extraction of quantitative texture features from medical images in an attempt to determine correlations with clinical endpoints. We hypothesize that convolutional ...neural networks (CNNs) could enhance the performance of traditional radiomics, by detecting image patterns that may not be covered by a traditional radiomic framework. We test this hypothesis by training a CNN to predict treatment outcomes of patients with head and neck squamous cell carcinoma, based solely on their pre-treatment computed tomography image. The training (194 patients) and validation sets (106 patients), which are mutually independent and include 4 institutions, come from The Cancer Imaging Archive. When compared to a traditional radiomic framework applied to the same patient cohort, our method results in a AUC of 0.88 in predicting distant metastasis. When combining our model with the previous model, the AUC improves to 0.92. Our framework yields models that are shown to explicitly recognize traditional radiomic features, be directly visualized and perform accurate outcome prediction.
To test whether altered radiation fractionation schemes (hyperfractionation HFX, accelerated fractionation, continuous AFX-C, and accelerated fractionation with split AFX-S) improved local-regional ...control (LRC) rates for patients with squamous cell cancers (SCC) of the head and neck when compared with standard fractionation (SFX) of 70 Gy.
Patients with stage III or IV (or stage II base of tongue) SCC (n=1076) were randomized to 4 treatment arms: (1) SFX, 70 Gy/35 daily fractions/7 weeks; (2) HFX, 81.6 Gy/68 twice-daily fractions/7 weeks; (3) AFX-S, 67.2 Gy/42 fractions/6 weeks with a 2-week rest after 38.4 Gy; and (4) AFX-C, 72 Gy/42 fractions/6 weeks. The 3 experimental arms were to be compared with SFX.
With patients censored for LRC at 5 years, only the comparison of HFX with SFX was significantly different: HFX, hazard ratio (HR) 0.79 (95% confidence interval 0.62-1.00), P=.05; AFX-C, 0.82 (95% confidence interval 0.65-1.05), P=.11. With patients censored at 5 years, HFX improved overall survival (HR 0.81, P=.05). Prevalence of any grade 3, 4, or 5 toxicity at 5 years; any feeding tube use after 180 days; or feeding tube use at 1 year did not differ significantly when the experimental arms were compared with SFX. When 7-week treatments were compared with 6-week treatments, accelerated fractionation appeared to increase grade 3, 4 or 5 toxicity at 5 years (P=.06). When the worst toxicity per patient was considered by treatment only, the AFX-C arm seemed to trend worse than the SFX arm when grade 0-2 was compared with grade 3-5 toxicity (P=.09).
At 5 years, only HFX improved LRC and overall survival for patients with locally advanced SCC without increasing late toxicity.
Although p16 protein expression, a surrogate marker of oncogenic human papillomavirus (HPV) infection, is recognized as a prognostic marker in oropharyngeal squamous cell carcinoma (OPSCC), its ...prevalence and significance have not been well established in cancer of the oral cavity, hypopharynx, or larynx, collectively referred as non-OPSCC, where HPV infection is less common than in the oropharynx.
p16 expression and high-risk HPV status in non-OPSCCs from RTOG 0129, 0234, and 0522 studies were determined by immunohistochemistry (IHC) and in situ hybridization (ISH). Hazard ratios from Cox models were expressed as positive or negative, stratified by trial, and adjusted for clinical characteristics.
p16 expression was positive in 14.1% (12 of 85), 24.2% (23 of 95), and 19.0% (27 of 142) and HPV ISH was positive in 6.5% (six of 93), 14.6% (15 of 103), and 6.9% (seven of 101) of non-OPSCCs from RTOG 0129, 0234, and 0522 studies, respectively. Hazard ratios for p16 expression were 0.63 (95% CI, 0.42 to 0.95; P = .03) and 0.56 (95% CI, 0.35 to 0.89; P = .01) for progression-free (PFS) and overall survival (OS), respectively. Comparing OPSCC and non-OPSCC, patients with p16-positive OPSCC have better PFS and OS than patients with p16-positive non-OPSCC, but patients with p16-negative OPSCC and non-OPSCC have similar outcomes.
Similar to results in patients with OPSCC, patients with p16-negative non-OPSCC have worse outcomes than patients with p16-positive non-OPSCC, and HPV may also have a role in outcome in a subset of non-OPSCC. However, further development of a p16 IHC scoring system in non-OPSCC and improvement of HPV detection methods are warranted before broad application in the clinical setting.
The presence of multiple serial organs at risk (OARs) in close proximity to the tumor makes treatment planning for glioblastoma (GBM) complex and time consuming. The present study aimed to create a ...knowledge-based (KB) radiation therapy model for GBM patients using RapidPlan.
An initial model was trained using 82 glioblastoma patients treated with 60 Gy in 30 fractions. Plans were created using either volumetric modulated arc therapy (VMAT) or intensity modulated radiation therapy (IMRT). To improve the goodness-of-fit of the model, an intermediate model was generated by using the dose-volume histograms (DVHs) of best spared OARs of the initial model. Using the intermediate model and manual refinement, all 82 cases were replanned, resulting in the final model. The final model was validated on an independent set of 45 patients with GBM, astrocytoma, oligodendroglioma, and meningioma.
The plans created by the final model exhibited superior planning target volume (PTV) dose metrics compared with manual clinical plans: ΔD
=-0.52 ± 0.20 Gy, and ΔD
=0.80 ± 0.13 Gy (differences are computed as clinical-model). OAR maximum doses were statistically similar, with improved optic apparatus sparing (ΔD
=2.78 ± 0.82 Gy). Stated improvements correspond to P<.05. The KB planning time is typically 7 minutes for IMRT and 13 minutes for VMAT, compared with a typical 4 hours for manual planning.
The KB approach results in significant improvement in planning efficiency and in superior PTV coverage and better normal tissue sparing irrespective of tumor size and location within the brain.
To evaluate the severity of cetuximab-induced skin rash and its correlation with clinical outcome and late skin toxicity in patients with head and neck squamous cell carcinoma treated with ...chemoradiation therapy and cetuximab.
Analysis included patients who received loading dose and ≥1 cetuximab dose concurrent with definitive chemoradiation therapy (70 Gy + cisplatin) or postoperative chemoradiation therapy (60-66 Gy + docetaxel or cisplatin).
Six hundred two patients were analyzed; 383 (63.6%) developed grade 2 to 4 cetuximab rash. Patients manifesting grade 2 to 4 rash had younger age (P<.001), fewer pack-years smoking history (P<.001), were more likely to be males (P=.04), and had p16-negative (P=.04) oropharyngeal tumors (P=.003). In univariate analysis, grade 2 to 4 rash was associated with better overall survival (hazard ratio HR 0.58, P<.001) and progression-free survival (HR 0.75, P=.02), and reduced distant metastasis rate (HR 0.61, P=.03), but not local-regional failure (HR 0.79, P=.16) relative to grade 0 to 1 rash. In multivariable analysis, HRs for overall survival, progression-free survival, distant metastasis, and local-regional failure were, respectively, 0.68 (P=.008), 0.85 (P=.21), 0.64 (P=.06), and 0.89 (P=.48). Grade ≥2 rash was associated with improved survival in p16-negative patients (HR 0.28 95% confidence interval 0.11-0.74) but not in p16-positive patients (HR 1.10 0.42-2.89) (P=.05 for interaction). Twenty-five percent of patients with grade 2 to 4 acute in-field radiation dermatitis experienced grade 2 to 4 late skin fibrosis, versus 14% of patients with grade 0 to 1 acute in-field radiation dermatitis (P=.002).
Grade 2 to 4 cetuximab rash was associated with better survival, possibly due to reduction of distant metastasis. This observation was noted mainly in p16-negative patients. Grade 2 to 4 acute in-field radiation dermatitis was associated with higher rate of late grade 2 to 4 skin fibrosis.
We performed a phase 2 trial of neoadjuvant temozolomide (TMZ), followed by hypofractionated accelerated radiation therapy (HART) with concurrent TMZ, and adjuvant TMZ in patients with newly ...diagnosed glioblastoma to determine whether neoadjuvant TMZ would safely improve outcomes in this group of patients prior to subsequent cytotoxic therapy.
Adult patients with newly diagnosed glioblastoma and a Karnofsky Performance Status >60 were eligible. Neoadjuvant TMZ administration started 2 to 3 weeks from surgery at a daily dose of 75 mg/m
for 2 weeks prior to delivery of HART (60 Gy in 20 daily fractions) with concurrent and adjuvant TMZ. The primary endpoints were feasibility and toxicity. The secondary endpoints included overall survival (OS) and progression-free survival.
Fifty patients were accrued. The median follow-up period was 44.0 months for patients at risk and 22.3 months for all 50 patients. Except for 1 patient in whom infection developed and another patient with progression during HART, all patients completed protocol therapy as planned. The median OS and progression-free survival were 22.3 months (95% confidence interval, 14.6-42.7 months) and 13.7 months (95% confidence interval, 8.0-33.3 months), respectively. The 4-year OS rates were 30.4% for the entire cohort and 53.3% and 14.0% for patients with methylated (n=21) and unmethylated (n=27) MGMT gene promoter tumors, respectively. One patient had grade 5 pancytopenia during HART, and another patient had transient grade 4 hepatotoxicity. A second surgical procedure was performed in 13 patients: 2 had intracranial infection, 3 had recurrences, 4 had recurrences and radiation-induced damage, and 4 had only radiation-induced damage.
This novel approach of neoadjuvant TMZ is associated with an encouraging favorable long-term survival with acceptable toxicity. A future comparative trial of the efficacy of this regimen is warranted.
To report long-term outcomes of low- and intermediate-risk prostate cancer patients treated with high-dose hypofractionated radiation therapy (HypoRT).
Patients with low- and intermediate-risk ...prostate cancer were treated using 3-dimensional conformal radiation therapy to a dose of 66 Gy in 22 daily fractions of 3 Gy without hormonal therapy. A uniform 7-mm margin was created around the prostate for the planning target volume, and treatment was prescribed to the isocenter. Treatment was delivered using daily ultrasound image-guided radiation therapy. Common Terminology Criteria for Adverse Events, version 3.0, was used to prospectively score toxicity. Biochemical failure was defined as the nadir prostate-specific antigen level plus 2 ng/mL.
A total of 129 patients were treated between November 2002 and December 2005. With a median follow-up of 90 months, the 5- and 8-year actuarial biochemical control rates were 97% and 92%, respectively. The 5- and 8-year actuarial overall survival rates were 92% and 88%, respectively. Only 1 patient died from prostate cancer at 92 months after treatment, giving an 8-year actuarial cancer-specific survival of 98%. Radiation therapy was well tolerated, with 57% of patients not experiencing any acute gastrointestinal (GI) or genitourinary (GU) toxicity. For late toxicity, the worst grade ≥2 rate for GI and GU toxicity was 27% and 33%, respectively. There was no grade >3 toxicity. At last follow-up, the rate of grade ≥2 for both GI and GU toxicity was only 1.5%.
Hypofractionation with 66 Gy in 22 fractions prescribed to the isocenter using 3-dimensional conformal radiation therapy produces excellent biochemical control rates, with moderate toxicity. However, this regimen cannot be extrapolated to the intensity modulated radiation therapy technique.
Background: Isotretinoin (13-cis-retinoic acid) is a synthetic vitamin A derivative, or retinoid, widely used in the treatment of cystic acne. Preclinical and clinical studies of high-dose ...isotretinoin in patients with head and neck squamous cell cancer (HNSCC) have produced encouraging results. We conducted a phase III randomized trial of low-dose isotretinoin versus placebo in early-stage HNSCC patients to assess its effect on second primary tumor incidence and survival. Methods: We randomly assigned 1190 patients who had been treated for stage I or II HNSCC to receive either low-dose isotretinoin (30 mg/day) or placebo for 3 years. The patients were monitored for up to 4 more years. Survival was analyzed by the Kaplan–Meier method, and Cox proportional hazards models were used for multivariable survival analysis. All statistical tests were two-sided. Results: Isotretinoin did not statistically significantly reduce the rate of second primary tumors (hazard ratio HR = 1.06, 95% confidence interval CI = 0.83 to 1.35) or increase survival (HR = 1.03, 95% CI = 0.81 to 1.32) compared with placebo in patients with early-stage HNSCC. Current smokers had a higher rate of second primary tumors than that of never (HR = 1.64, 95% CI = 1.08 to 2.50) or former (HR = 1.32, 95% CI = 1.01 to 1.71) smokers. The hazard ratio of death from any cause for current smokers versus never smokers was 2.51 (95% CI = 1.54 to 4.10) and for current smokers versus former smokers was 1.60 (95% CI = 1.23 to 2.07). Major sites of second primary tumors (n = 261) included lung (31%), oral cavity (17%), larynx (8%), and pharynx (5%). Conclusions: Low-dose isotretinoin was not effective in reducing the rate of second primary tumors or death or smoking-related disease. Smoking statistically significantly increased the rate of second primary tumors and death. Ongoing trials are testing higher doses of isotretinoin as part of combination bioadjuvant therapeutic methods for patients with locally advanced HNSCC.
To analyze the quality of life (QOL) and performance status (PS) (secondary outcome) in patients with stage III to IV head and neck cancer (HNC) enrolled on a prospective randomized phase 3 trial ...comparing radiation-cisplatin without cetuximab (CIS) or with cetuximab (CET/CIS). The QOL hypothesis proposed a between-arm difference in Functional Assessment of Cancer Therapy-Head and Neck (FACT-HN) total score of ≥10% of the instrument range from baseline to 1 year.
Patients who gave consent to the QOL/PS study completed the FACT-HN, Performance Status Scale for HNC (PSS-HN), and EuroQol (EQ-5D) at baseline through to 5 years. The pretreatment QOL/PS scores were correlated with outcome and p16 status in patients with oropharyngeal cancer (OPC).
Of 818 analyzable patients, the 1-year change from baseline score for FACT-HN total was -0.41 (CIS arm) and -5.11 (CET/CIS arm) (P=.016), representing a 3.2% between-arm change of the FACT-HN total score. The mean EQ-5D index and PSS-HN scores were not significantly different between arms. The p16-positive OPC patients had significantly higher baseline and 1-year scores for PSS-HN, FACT-HN total, physical and functional subscales, and 2-years for the EQ-5D index compared with p16-negative OPC patients. Higher pretreatment PSS-HN diet, PSS-HN eating, FACT-HN, and EQ-5D index scores were associated with better overall survival (OS) and progression-free (PFS) survival on multivariate analysis. Higher baseline FACT-HN total, functional, physical subscale, and EQ-5D index scores were associated with improved OS and PFS in p16-positive OPC patients but not in p16-negative and non-OPC patients.
There was no clinically meaningful difference in QOL/PS between arms. The p16-positive OPC patients had significantly higher QOL/PS than did p16-negative patients. Pretreatment QOL/PS is a significant independent predictor of outcome in locally advanced HNC.