Calibration of polygenic risk scores is required prior to clinical implementation: results of three common cancers in UKB.

BACKGROUND: SNP-based polygenic risk scores have recently been adopted in the clinic for risk assessment of some common diseases. Their validity is supported by a consistent trend between their percentile rank and disease risk in populations. However, for clinical use at the individual level, the reliability of score values is necessary considering they are directly used to calculate remaining lifetime risk. OBJECTIVES: We assessed the reliability of polygenic score values to estimate prostate cancer (PCa), breast cancer (BCa) and colorectal cancer (CRC) risk in three incident cohorts from the UK Biobank (n>500 000). METHODS: Cancer-specific Genetic Risk Score (GRS), a well-established population-standardised polygenic risk score, was calculated. RESULTS: A systematic bias was found between estimated risks (GRS values) and observed risks; ß (95% CI) was 0.67 (0.58-0.76), 0.74 (0.65-0.84) and 0.82 (0.75-0.89), respectively, for PCa, BCa and CRC, all significantly lower than 1.00 (perfect calibration), p<0.001. After applying a correction factor derived from a training data set, the ß for corrected GRS values in an independent testing data set were 1.09 (1.05-1.13), 1.00 (0.88-1.12) and 1.08 (0.96-1.21), respectively, for PCa, BCa and CRC. CONCLUSION: Assessing the calibration of polygenic risk scores is necessary and feasible to ensure their reliability prior to clinical implementation.

A tool to predict disparities in the timeliness of surgical treatment for breast cancer patients in the USA.

PURPOSE: Breast cancer outcomes are impaired by both delays and disparities in treatment. This study was performed to assess their relationship and to provide a tool to predict patient socioeconomic factors associated with risk for delay. METHODS: The National Cancer Database was reviewed between 2004 and 2017 for patients with non-metastatic breast cancer managed with upfront surgery. Times to treatment were measured from the date of diagnosis. Patient, tumor, and treatment factors were assessed with attention paid to sociodemographic variables. RESULTS: 514,187 patients remained after exclusions, with 84.3% White, 10.8% Black, 3.7% Asian, and Hispanics comprising 5.6% of the cohort. Medicaid and uninsured patients had longer mean adjusted time to surgery (≥ 46 days) versus private (36.7 days), Medicare (35.9 days), or other governmental insurance (39.8 days). After adjustment, Black race and Hispanic ethnicity were most impactful, adding 6.0 and 6.4 preoperative days, 10.9 and 11.5 days to chemotherapy, 11.1 and 9.1 days to radiation, and 12.5 and 8.9 days to endocrine therapy, respectively. Income, education, and insurance, among other factors, also affected delay. A nomogram, including race and sociodemographic factors, was created to predict the risk of preoperative delay. CONCLUSION:  Significant disparities exist in timeliness of care for factors, including but not limited to, race and ethnicity. Although exact causes cannot be discerned, these data indicate population subsets whose intervals of care risk being longer than those specified by national quality standards. The nomogram created here may help direct resources to those at highest risk of incurring a treatment delay.

The Impact of a Machine Learning Early Warning Score on Hospital Mortality: A Multicenter Clinical Intervention Trial.

OBJECTIVES: To determine the impact of a machine learning early warning risk score, electronic Cardiac Arrest Risk Triage (eCART), on mortality for elevated-risk adult inpatients. DESIGN: A pragmatic pre- and post-intervention study conducted over the same 10-month period in 2 consecutive years. SETTING: Four-hospital community-academic health system. PATIENTS: All adult patients admitted to a medical-surgical ward. INTERVENTIONS: During the baseline period, clinicians were blinded to eCART scores. During the intervention period, scores were presented to providers. Scores greater than or equal to 95th percentile were designated high risk prompting a physician assessment for ICU admission. Scores between the 89th and 95th percentiles were designated intermediate risk, triggering a nurse-directed workflow that included measuring vital signs every 2 hours and contacting a physician to review the treatment plan. MEASUREMENTS AND MAIN RESULTS: The primary outcome was all-cause inhospital mortality. Secondary measures included vital sign assessment within 2 hours, ICU transfer rate, and time to ICU transfer. A total of 60,261 patients were admitted during the study period, of which 6,681 (11.1%) met inclusion criteria (baseline period n = 3,191, intervention period n = 3,490). The intervention period was associated with a significant decrease in hospital mortality for the main cohort (8.8% vs 13.9%; p < 0.0001; adjusted odds ratio [OR], 0.60 [95% CI, 0.52-0.71]). A significant decrease in mortality was also seen for the average-risk cohort not subject to the intervention (0.49% vs 0.26%; p < 0.05; adjusted OR, 0.53 [95% CI, 0.41-0.74]). In subgroup analysis, the benefit was seen in both high- (17.9% vs 23.9%; p = 0.001) and intermediate-risk (2.0% vs 4.0 %; p = 0.005) patients. The intervention period was also associated with a significant increase in ICU transfers, decrease in time to ICU transfer, and increase in vital sign reassessment within 2 hours. CONCLUSIONS: Implementation of a machine learning early warning score-driven protocol was associated with reduced inhospital mortality, likely driven by earlier and more frequent ICU transfer.

Association of prostate cancer polygenic risk score with number and laterality of tumor cores in active surveillance patients.

BACKGROUND: Prostate cancer (PCa) is characterized by its tendency to be multifocal. However, few studies have investigated the endogenous factors that explain the multifocal disease. The primary objective of the current study is to test whether inherited PCa risk is associated with multifocal tumors in PCa patients. METHODS: Subjects in this study were PCa patients of European ancestry undergoing active surveillance at Johns Hopkins Hospital (N = 805) and NorthShore University HealthSystem (N = 432). The inherited risk was measured by genetic risk score (GRS), an odds ratio-weighted and population-standardized polygenic risk score based on known risk-associated single nucleotide polymorphisms. PCa multifocality was indirectly measured by the number and laterality of positive tumor cores from a 12-core systematic biopsy. RESULTS: In the combined cohort, 35.7% and 66.3% of patients had ≥2 tumor cores at the initial diagnostic biopsy and on at least one subsequent surveillance biopsy, respectively. For tumor laterality, 7.8% and 47.8% of patients had bilateral tumor cores at diagnostic and surveillance biopsies, respectively. We found, for the first time, that patients with higher numbers of positive cores at diagnostic and surveillance biopsies, respectively, had significantly higher mean GRS values; p = .01 and p = 5.94E-04. Additionally, patients with bilateral tumors at diagnostic and surveillance biopsies, respectively, had significantly higher mean GRS values than those with unilateral tumors; p = .04 and p = .01. In contrast, no association was found between GRS and maximum core length of tumor or tumor grade at diagnostic/surveillance biopsies (all p > .05). Finally, we observed a modest trend that patients with higher GRS quartiles had a higher risk for tumor upgrading on surveillance biopsies. The trend, however, was not statistically significant (p > .05). CONCLUSIONS: The associations of GRS with two measurements of PCa multifocality (core numbers and laterality) provide novel and consistent evidence for the link between inherited PCa risk and multifocal tumors.

Feasibility and performance of a novel probe panel to detect somatic DNA copy number alterations in clinical specimens for predicting prostate cancer progression.

BACKGROUND: To assess the feasibility of a novel DNA-based probe panel to detect copy number alterations (CNAs) in prostate tumor DNA and its performance for predicting clinical progression. METHODS: A probe panel was developed and optimized to measure CNAs in trace amounts of tumor DNA (2 ng) isolated from formalin-fixed paraffin-embedded tissues. Ten genes previously associated with aggressive disease were targeted. The panel's feasibility and performance were assessed in 175 prostate cancer (PCa) patients who underwent radical prostatectomy with a median 10-year follow-up, including 42 men who developed disease progression (either metastasis and/or PCa-specific death). Association with disease progression was tested using univariable and multivariable analyses. RESULTS: The probe panel detected CNAs in all 10 genes in tumor DNA isolated from either diagnostic biopsies or surgical specimens. A four-gene model (PTEN/MYC/BRCA2/CDKN1B) had the strongest association with disease progression; 64.3% of progressors and 22.5% of non-progressors had at least one CNA in these four genes, odds ratio (OR) (95% confidence interval) = 6.21 (2.77-13.87), P = 8.48E-06. The association with disease progression remained significant after adjusting for known clinicopathological variables. Among the seven progressors of the 65 patients with clinically low-risk disease, three (42.9%) had at least one CNA in these four genes. CONCLUSIONS: The probe panel can detect CNAs in trace amounts of tumor DNA from biopsies or surgical tissues at the time of diagnosis or surgery. CNAs independently predict metastatic/lethal cancer, particularly among men with clinically low-risk disease at diagnosis. If validated, this may improve current abilities to assess tumor aggressiveness.

Germline HOXB13 G84E mutation carriers and risk to twenty common types of cancer: results from the UK Biobank.

Germline HOXB13 G84E mutation has been consistently associated with prostate cancer (PCa) risk, but its association with other cancers is controversial. We systematically tested its association with the 20 most common cancer types in subjects from the UK Biobank. The G84E mutation was found in 1,545 (0.34%) of 460,224 participants of European ancestry. While mutation status did not associate with cancer risk in females, it was significantly associated with increased risk in males; odds ratio (OR) (95% confidence interval) for overall cancer diagnosis was 2.19 (1.89-2.52), P = 2.5E-19. The association remained after excluding PCa; OR = 1.4 (1.16-1.68), P = 0.003, suggesting association with other cancers. Indeed, suggestive novel associations were found for two other cancer types; rectosigmoid cancer, OR = 2.25 (1.05-4.15), P = 0.05 and non-melanoma skin cancer (NMSC), OR = 1.40 (1.12-1.74), P = 0.01. For NMSC, the association was found only in basal cell carcinoma, OR = 1.37 (1.07-1.74), P = 0.03. These findings have potential clinical utility for genetic counselling regarding HOXB13.

Concept and benchmarks for assessing narrow-sense validity of genetic risk score values.

BACKGROUND: While higher genetic risk score (GRS) has been statistically associated with increased disease risk (broad-sense validity), the concept and tools for assessing the validity of reported GRS values from tests (narrow-sense validity) are underdeveloped. METHODS: We propose two benchmarks for assessing the narrow-sense validity of GRS. The baseline benchmark requires that the mean GRS value in a general population approximates 1.0. The calibration benchmark assesses the agreement between observed risks and estimated risks (GRS values). We assessed benchmark performance for three prostate cancer (PCa) GRS tests, derived from three SNP panels with increasing stringency of selection criteria, in a PCa chemoprevention trial where 714 of 3225 men were diagnosed with PCa during the 4-year follow-up. RESULTS: GRS from Panels 1, 2, and 3 were all statistically associated with PCa risk; P = 5.58 × 10-3 , P = 1 × 10-3 , and P = 1.5 × 10-13 , respectively (broad-sense validity). For narrow-sense validity, the mean GRS value among men without PCa was 1.33, 1.09, and 0.98 for Panels 1, 2, and 3, respectively (baseline benchmark). For assessing the calibration benchmark, observed risks were calculated for seven groups of men with GRS values <0.3, 0.3-0.79, 0.8-1.19, 1.2-1.49, 1.5-1.99, 2-2.99, and ≥3. The calibration slope (higher is better) was 0.15, 0.12, and 0.60, and the bias score (lower is better) between the observed risks and GRS values was 0.08, 0.08, and 0.02 for Panels 1, 2, and 3, respectively. CONCLUSION: Performance differed considerably among GRS tests. We recommend that all GRS tests be evaluated using the two benchmarks before clinical implementation for individual risk assessment.

Treatment delays from transfers of care and their impact on breast cancer quality measures.

PURPOSE: Despite delays between diagnosis and surgery adversely affecting survival, patients frequently transfer their breast cancer care between institutions. This study was performed to assess the prevalence and effect of such transfers of care (TsOC) on the time to surgery, and its impact on current time-dependent breast cancer quality metrics at Commission on Cancer (CoC) and National Accreditation Program for Breast Centers (NAPBC)-accredited institutions. METHODS: Patients having non-metastatic invasive breast cancer diagnosed between 2006 and 2015 at CoC and NAPBC centers ("reporting facilities") in the National Cancer Database were reviewed. TsOC refer to transferring into or out of a reporting facility between diagnosis and surgery. RESULTS: Among 622,793 patients, 36.6% of patients transferred care. TsOC add 7.3, 7.8, 8.7, and 9.8 days in time to surgery, chemotherapy, radiotherapy, and endocrine therapy, respectively (p's < 0.0001). On multivariable analysis, the odds of surgery occurring > 90 days from diagnosis were greatest for patients undergoing unilateral or bilateral mastectomy, Black or Hispanic patients, and those having TsOC (ORs > 1.73, p's < 0.0001). TsOC increase the odds of non-compliance, per patient, for chemotherapy, radiotherapy and endocrine therapy time-dependent measures by 65.4%, 25.6%, and 56.5%, respectively (p < 0.0001). CONCLUSIONS: TsOC for newly diagnosed breast cancers to or from an accredited facility result in delays in time to surgery which can affect compliance with time-dependent quality measures. Facilities frequently receiving transferred patients may be most adversely affected. Although non-compliance with these quality measures is low, institutions and accrediting bodies should be aware of these associations in order to comply with time-dependent standards.

Impact of Breast Center Accreditation on Compliance with Breast Quality Performance Measures at Commission on Cancer-Accredited Centers.

PURPOSE: This study was designed to determine whether accreditation by the National Accreditation Program for Breast Centers (NAPBC) is associated with improved performance on six breast quality measures pertaining to adjuvant treatment, needle/core biopsy, and breast conservation therapy rates at Commission on Cancer (CoC) centers. METHODS: National Cancer Database 2015 data were retrospectively reviewed to compare patients treated at CoC centers with and without NAPBC accreditation for compliance on six breast cancer quality measures. Mixed effects modeling determined performance on the quality measures adjusting for patient, tumor, and facility factors. RESULTS: Of 1308 CoC facilities, 484 (37%) were NAPBC-accredited and 111,547 patients (48%) were treated at NAPBC centers. More than 80% of patients treated at both NAPBC and non-NAPBC centers received care in compliance with breast quality measures. NAPBC centers achieved significantly higher performance on four of the five quality measures than non-NAPBC centers at the patient level and on five of six measures at the facility level. For two measures, needle/core biopsy before surgical treatment of breast cancer and breast conservation therapy rate of 50%, NAPBC centers were twice as likely as non-NAPBC centers to perform at the level expected by the CoC (respectively odds ratio [OR] 1.96, 95% confidence interval [CI] 1.85-2.08, p < 0.0001; and OR 2.05, 95% CI 1.94-2.15, p < 0.0001). CONCLUSIONS: While NAPBC accreditation at CoC centers is associated with higher performance on breast quality measures, the majority of patients at all centers receive guideline-concordant care. Future studies will determine whether higher performance translates into improved oncologic and patient-reported outcomes.

Genetic risk score modifies the effect of APOE on risk and age onset of Alzheimer's disease.

Single nucleotide polymorphism (SNP)-based genetic risk score (GRS) and APOE genotype are both important in risk prediction of Alzheimer's disease (AD); however, the interaction between GRS and APOE has not been extensively investigated. Our objective was to determine whether GRS modifies the APOE effect on AD risk and age at onset (AAO). The study included 774 AD cases and 767 controls of European descent. Population standardized GRS was calculated based on 17 previously implicated AD risk-associated SNPs. Association was analyzed using logistic regression, Cox proportional hazards model and Kaplan-Meier curve. We found that GRS was significantly associated with AD risk and the association was stronger among APOE ε4 carriers. Compared to ε4 non-carriers, the Odds Ratio (OR) for AD was 8.09 (95% Confidence Interval [CI]: 4.98-13.63) for ε4 carriers with high-GRS (≥1.5). In contrast, the OR was 2.55 (95% CI: 1.46-4.49) for ε4 carriers with low-GRS (<0.6). In conclusion, these results suggest SNP-based GRS may supplement APOE for better assessment of inherited risk and age of onset of AD.

Regional Variation in Performance for Commission on Cancer Breast Quality Measures and Impact on Overall Survival.

BACKGROUND: Adherence to quality measures has become an important indicator of cancer center performance for high-quality cancer care. We examined regional variation in performance for Commission on Cancer breast quality measures and its impact on overall survival (OS) for those measures that have been shown to impact OS. METHODS: Six breast quality measures were analyzed using the National Cancer Data Base from 2014 to 2015, and a multivariable model was used to assess performance for each measure by region. Kaplan-Meier and Cox proportional hazard models were used to examine OS between high- and low-performing centers from 2007 to 2012. RESULTS: Overall, 305,391 women had surgery at 1322 institutions in nine US regions; 90.8% underwent needle biopsy (range 86.0-92.6% between regions, p < 0.01), 69.8% had breast-conserving surgery (BCS) for stage 0-II cancer (60.9-79.3%, p < 0.01), 85.2% aged < 70 years had radiation therapy (RT) after BCS (79.6-90.8%, p < 0.01), 78.3% of women with four or more positive nodes had post-mastectomy RT (70.9-84.5%, p < 0.01), 90.9% with hormone receptor (HR)-positive stage IB-III cancer had hormone therapy (83.7-95.1%, p < 0.01), and 89.4% aged < 70 years with HR-negative stage IB-III cancer had chemotherapy (87.6-91.4%, p < 0.01). Multivariate analyses adjusted for patient and facility factors found that region was the only consistent predictor of non-compliance across measures. With median 65-month follow-up, there was no difference in OS between high- and low-performing centers for the three measures that have been shown to impact OS. CONCLUSIONS: There is significant regional variation in performance on the breast quality measures but this variation did not impact OS. Targeted efforts in certain areas of the country may help improve performance on these measures.

The Shifting Paradigm for Breast Cancer Surgery in Patients Undergoing Neoadjuvant Chemotherapy.

INTRODUCTION: Surgical therapy for newly diagnosed breast cancer has changed over the past decade, but these trends have not been well documented in patients undergoing neoadjuvant therapy (NAC). METHODS: In a retrospective cohort study of the National Cancer Database (NCDB), we selected 285,514 women with clinical stage I-III breast cancer who underwent NAC or adjuvant therapy (AC) from 2006 to 2014. Breast-conserving surgery (BCS), unilateral mastectomy (UM), and bilateral mastectomy (BM) rates were compared between patients undergoing NAC and AC. RESULTS: Of 285,514 women, 68,850 (24.1%) underwent NAC. Of NAC patients, 18,158 (26.4%) underwent BM and 27,349 (39.7%) BCS compared with 31,886 (14.7%) and 120,626 (55.7%) AC patients, respectively. From 2006 to 2014, BM increased from 16.1 to 28.8% (p < 0.001) for NAC and from 7.4 to 17.5% (p < 0.001) for AC. After adjusting for patient, tumor, and facility factors, NAC patients were 1.50 times [odds ratio (OR) 1.50, confidence interval (CI) 1.42-1.51] more likely to undergo BM then AC patients. The difference in BM rates between patients receiving NAC versus AC varied significantly by cT classification. This difference was the greatest among cT1 tumors between NAC and AC (31.7 vs. 13.0%, p < 0.001), followed by cT2 tumors (24.1 vs. 16.6%, p < 0.001) and cT3 tumors (24.3 vs. 22.3%). CONCLUSIONS AND RELEVANCE: More NAC patients are undergoing BM while fewer are undergoing BCS compared with patients undergoing AC. This trend is particularly striking for those patients with smaller tumors who would otherwise be candidates for BCS.

Differences in the Impact of Age on Mortality in Well-Differentiated Thyroid Cancer.

INTRODUCTION: Well-differentiated thyroid cancer (WDTC) is unique in that patient age is part of staging. Several studies have shown a need to increase the age threshold in staging for WDTC, but the separate impact of age on prognosis for papillary and follicular carcinomas has not been examined. We hypothesize that age impacts survival differently for papillary and follicular carcinomas. METHODS: Patients with invasive papillary thyroid carcinoma (PTC) and follicular thyroid carcinoma (FTC) between 2004 and 2013 were identified in the National Cancer Database, and were stratified by histologic type. Overall survival (OS) was analyzed using multivariable Cox regression, and the Youden index was used to find the optimal age threshold for both histologies. RESULTS: A total of 204,139 patients with WDTC were identified. Ninety-two percent had PTC, while 7.7% had FTC. The average age was 48.4 years and OS was 96.3%, with a median follow-up of 52.7 months. When analyzing age in 5-year increments, 10-year mortality increased incrementally by 30-50% per age group for PTC, from age < 35 to ≥ 70 years, without an obvious inflection point. However, FTC patients experienced a more than threefold increase in 10-year mortality from age 40-44 years (2.5%) to age 45-49 years (7.9%). The same pattern was found on multivariable Cox regression. The Youden index found the optimal age thresholds were 58.5 years for PTC and 46.2 years for FTC. CONCLUSION: OS for PTC decreases incrementally with age, but OS for FTC decreases significantly in patients aged 45 years and older. A higher age threshold may inappropriately downstage some high-risk follicular cancer patients.

CONCORDANCE OF PRE-OPERATIVE CLINICAL STAGE WITH PATHOLOGIC STAGE IN PATIENTS ≥45 YEARS OLD WITH WELL-DIFFERENTIATED THYROID CANCER.

OBJECTIVE: Clinical stage (cStage) in thyroid cancer determines extent of surgical therapy and completeness of resection. Pathologic stage (pStage) is an important determinant of outcome. The rate of discordance between clinical and pathologic stage in thyroid cancer is unknown. METHODS: The National Cancer Data Base was queried to identify 27,473 patients ≥45 years old with cStage I through IV differentiated thyroid cancer undergoing surgery from 2008-2012. RESULTS: There were 16,286 (59.3%) cStage I patients; 4,825 (17.6%) cStage II; 4,329 (15.8%) cStage III; and 2,013 (7.3%) cStage IV patients. The upstage rate was 15.1%, and the downstage rate was 4.6%. For cStage II, there was a 25.5% upstage rate. The change in cStage was a result of inaccurate T-category in 40.8%, N-category in 36.3%, and both in 22.9%. On multivariate analysis, the patients more likely to be upstaged had papillary histology, tumors 2.1 to 4 cm, total thyroidectomy, nodal surgery, positive margins, or multifocal disease. Upstaged patients received radioiodine more frequently (75.3% vs. 48.1%; P<.001). CONCLUSION: Approximately 20% of cStage is discordant to pStage. Certain populations are at risk for inaccurate staging, including cT2 and cN0 patients. Upstaged patients are more likely to receive radioactive iodine therapy. ABBREVIATIONS: CI = confidence interval; cStage = clinical stage; DTC = differentiated thyroid cancer; NCDB = National Cancer Data Base; OR = odds ratio; pStage = pathologic stage; RAI = radioactive iodine.

The extent of vascular resection is associated with perioperative outcome in patients undergoing pancreaticoduodenectomy.

BACKGROUND: Few studies have examined the relation between extent of vascular resection and morbidity following pancreaticoduodenectomy (PD) with vein resection (PDVR). METHODS: Patients undergoing PD for malignancy were identified using the American College of Surgeons National Surgical Quality Improvement Project from 2006 to 2013. Current procedural terminology codes were used to characterize PDVR. RESULTS: 9235 patients underwent PD, 977 (10.6%) had PDVR - 640 with direct and 224 with graft repair. PDVR had longer operative times (456 ± 136 vs 374 ± 128 min, p < 0.05) and higher intraoperative transfusions (1.8 ± 3.4 vs 4.3 ± 4.9 units, p < 0.05) than PD alone. On adjusted multivariable regression, PDVR with either direct or graft repairs was associated with higher rates of overall morbidity (OR [odds ratio] 1.50 for direct, 1.74 for graft, p < 0.05), bleeding (OR 2.18 for direct, 3.26 for graft, p < 0.05), and DVT (OR 2.12 for direct, 2.62 for graft, p < 0.05) compared to PD alone. Graft repair was further associated with increased risk of reoperation (OR 1.59), septic shock (OR 2.77) and 30-day mortality (OR 2.72), all p < 0.05. DISCUSSION: The risk of significant morbidity and mortality for PDVR is associated with the extent of vascular resection, with graft repairs having increased morbidity and mortality rates.

The national landscape of human papillomavirus-associated oropharynx squamous cell carcinoma.

The head and neck squamous cell carcinoma (HNC) landscape is evolving with human papillomavirus (HPV) being a rising cause of oropharynx carcinoma (OPC). This study seeks to investigate a national database for HPV-associated oropharynx carcinoma (HPV-OPC). Using the National Cancer Data Base, we analyzed 22,693 patients with HPV-OPC and known HPV status. Chi-square tests and logistic regression models were utilized to examine differences between HPV positive and HPV negative OPC. 14,805 (65.2%) patients were HPV positive. Mean age at presentation was 58.4 years with HPV-HNC patients being 2.8 years younger compared to the HPV-negative cohort (58.4 vs. 61.2 years, p < 0.001). 67.6% of white patients were HPV-positive compared to 42.3% of African American patients and 57.1% of Hispanics (p < 0.001). When combining race and socioeconomic status (SES), we found African American patients in high SES groups had HPV-OPC prevalence that was significantly higher than African American patients in low SES groups (56.9% vs. 36.3%, p < 0.001). Geographic distribution of HPV-OPC was also analyzed and found to be most prevalent in Western states and least prevalent in the Southern states (p < 0.001). The distribution of HPV-OPC is variable across the country and among racial and socioeconomic groups. A broad understanding of these differences in HPV-OPC should drive educational programs and improve clinical trials that benefit both prevention and current treatments.

The Impact of Facility Volume on Rates of Pathologic Complete Response to Neoadjuvant Chemotherapy Used in Breast Cancer.

BACKGROUND: Patient and tumor factors have been associated with rates for pathologic complete response (pCR) to neoadjuvant chemotherapy (NAC) for breast cancer, but variation in pCR rates across facilities has not been studied. METHODS: This study used the National Cancer Data Base to identify women with clinical stages 1-3a breast cancer undergoing NAC from 2010 to 2013. Generalized estimation equation models were used to examine the relationship between facility characteristics and pCR rates, with adjustment for patient and tumor factors, while accounting for patient clustering at facilities. Analyses were stratified by tumor molecular subtype. RESULTS: Overall, 16,885 women underwent NAC, of whom 3130 (18.5%) were hormone receptor-positive (HR+) and human epidermal growth factor 2-positive (HER2+), 7045 (41.7%) were HR+HER2-, 1847 (10.9%) were HR-HER2+, and 4863 (28.8%) were HR-HER2-. Overall, 4002 of the patients (23.7%) achieved a pCR. The pCR rates were 29.5% for HR+HER2+, 10.8% for HR+HER2-, 45.3% for HR-HER2+, and 30.5% for HR-HER2- tumors. Multivariable analysis showed that pCR rates were significantly higher at high-volume facilities (>75th vs. <25th percentile) for all tumor subtypes except HR+HER2- tumors. Facility location and type were not significant. Adjustment for time from NAC to surgery decreased the likelihood of a pCR in high- versus low-volume facilities, but facility volume remained significantly associated with pCR. CONCLUSION: Facility volume, not location or type, was significantly associated with higher pCR rates in this exploratory analysis. Time to surgery has a modest impact on pCR rates across facilities, but further study to identify other potentially modifiable factors is needed.

Impact of an In-visit Decision Aid on Patient Knowledge about Contralateral Prophylactic Mastectomy: A Pilot Study.

BACKGROUND: Studies have reported that breast cancer patients have limited understanding about the oncologic outcomes following contralateral prophylactic mastectomy (CPM). We hypothesized that an in-visit decision aid (DA) would be associated with higher patient knowledge about the anticipated short and long term outcomes of CPM. METHODS: We piloted a DA which used the SCOPED: (Situation, Choices, Objectives, People, Evaluation and Decision) framework. Knowledge, dichotomized as "low" (≤3 correct) versus "high" (≥4 correct), was assessed immediately after the visit by a 5 item survey. There were 97 DA patients (response rate 62.2 %) and 114 usual care (UC) patients (response rate 71.3 %). RESULTS: Patient demographic factors were similar between the two groups. Twenty-one (21.7 %) patients in the DA group underwent CPM compared with 18 (15.8 %) in the UC group (p = 0.22). Mean and median knowledge levels were significantly higher in the DA group compared with the UC group for patients of all ages, tumor stage, race, family history, anxiety levels, worry about CBC, and surgery type. Eighty-six (78.9 %) of UC versus 35 (37.9 %) DA patients had low knowledge. Of patients who underwent CPM, 15 (83.3 %) in the UC cohort versus 5 (25.0 %) of DA patients had "low" knowledge. CONCLUSIONS: Knowledge was higher in the DA group. The UC group had approximately three times the number of patients of the DA group who were at risk for making a poorly informed decision to have CPM. Future studies should assess the impact of increased knowledge on overall CPM rates.

Post-mastectomy radiation therapy and overall survival after neoadjuvant chemotherapy.

BACKGROUND: The role of postmastectomy radiation therapy (PMRT) after neoadjuvant chemotherapy (NAC) and mastectomy is unclear, especially in patients that have post-treatment tumor negative axillary nodes (ypN0). METHODS: The National Cancer Data Base was used to identify women that had PMRT after NAC and mastectomy for clinically node positive (cN1-2) disease from 2004 to 2008. Median follow-up time was 69 months. RESULTS: 8,321 patients were included for analysis, and 6140 (65.6%) had cN1 disease and 2181 (23.3%) had cN2 disease. On adjusted survival analysis, PMRT was associated with an overall survival (OS) benefit in both patients with cN1 (5-yr OS 75.8% vs. 71.9%, P < 0.01) and cN2 (5-yr OS 69.2% vs. 58.6%, P < 0.01) disease. In the subgroup of patients that were ypN0 after NAC, there was no significant survival difference (P > 0.11) for PMRT compared to those patients who were not ypN0, except for patients with hormone-receptor negative tumors, who had improved OS with PMRT (HR 0.65, P < 0.01). CONCLUSIONS: PMRT is associated with improved OS in patients with cN1 and cN2 disease after NAC and mastectomy. However, in the subgroup of patients that were ypN0 after NAC, PMRT improved OS for hormone-receptor negative patients but not hormone-receptor positive patients.

Clinical accuracy of preoperative breast MRI for breast cancer.

BACKGROUND: It is unclear if breast magnetic resonance imaging (MRI) is more accurate than mammography (MGM) and ultrasound (U/S) in aggregate for patients with invasive cancer. METHODS: We compared concordance of combined tumor size and tumor foci between MRI and MGM and U/S combined to pathological tumor size and foci as the gold standard from 2009 to 2015. Tumor size was nonconcordant if it differed from the pathologic size by ≥33% and tumor foci was nonconcordant if >1 foci were seen. If one or both of the MGM or U/S was nonconcordant and the MRI was concordant, MRI provided greater accuracy. RESULTS: Of 471 patients with MGM, US, and MRI, MRI was more accurate for 32.9% of patients for tumor size and for 21.9% for tumor foci. Patients for whom MRI had greater accuracy were compared to those who did not for clinical and tumor factors. The only significant factor was calcifications on mammography. Tumor size, stage, molecular subtype, histology, grade, patient BMI, age, mammographic density, and use of hormone replacement therapy were not significantly different. CONCLUSIONS: Breast MRI provides greater accuracy for a third of patients undergoing preoperative MGM and U/S. Mammographic calcifications were associated with MRI clinical accuracy for patients with invasive cancer.