Cause-specific Research Articles

Causal inference and counterfactual prediction in machine learning for actionable healthcare

Big data, high-performance computing, and (deep) machine learning are increasingly becoming key to precision medicine—from identifying disease risks and taking preventive measures, to making diagnoses and personalizing treatment for individuals. Precision medicine, however, is not only about predicting risks and outcomes, but also about weighing interventions. Interventional clinical predictive models require the correct specification of cause and effect, and the calculation of so-called counterfactuals, that is, alternative scenarios. In biomedical research, observational studies are commonly affected by confounding and selection bias. Without robust assumptions, often requiring a priori domain knowledge, causal inference is not feasible. Data-driven prediction models are often mistakenly used to draw causal effects, but neither their parameters nor their predictions necessarily have a causal interpretation. Therefore, the premise that data-driven prediction models lead to trustable decisions/interventions for precision medicine is questionable. When pursuing intervention modelling, the bio-health informatics community needs to employ causal approaches and learn causal structures. Here we discuss how target trials (algorithmic emulation of randomized studies), transportability (the licence to transfer causal effects from one population to another) and prediction invariance (where a true causal model is contained in the set of all prediction models whose accuracy does not vary across different settings) are linchpins to developing and testing intervention models. Machine learning models are commonly used to predict risks and outcomes in biomedical research. But healthcare often requires information about cause–effect relations and alternative scenarios, that is, counterfactuals. Prosperi et al. discuss the importance of interventional and counterfactual models, as opposed to purely predictive models, in the context of precision medicine.

Incidence and outcomes of neuroblastoma in Australian children: A population-based study (1983-2015).

Neuroblastoma predominantly affects younger children and exhibits heterogeneous behaviour. This study describes incidence and outcomes for neuroblastoma using national population-based data from the Australian Childhood Cancer Registry. Deidentified data for all children (0-14 years) diagnosed with neuroblastoma and ganglioneuroblastoma from 1983 to 2015 were extracted. Cause-specific (CSS) and event-free survival were estimated using the cohort method. Adjusted hazard ratios were calculated using a multivariable flexible parametric survival model. Other outcomes investigated included recurrence and second primary malignancies (SPMs). The study cohort comprised 1269 patients. Age-standardised incidence rates remained steady across the study period at approximately 9.5 per million children per year. The proportion of patients with metastatic disease at diagnosis decreased from 63% in 1983-1995 to 42% by 2006-2015 (P < 0.001). CSS and event-free survival both improved significantly over time and reached 75% (95% confidence interval (CI) = 71-79%) and 71% (95% CI = 66-75%) at 5 years post-diagnosis, respectively, for children diagnosed between 2004 and 2013. Of patients achieving full remission, 28% relapsed with subsequent 5-year CSS of only 20%. Although SPMs were rare, neuroblastoma survivors carried a fivefold increased risk compared to cancer rates in the general population (standardised incidence ratio = 5.18, 95% CI = 3.01-8.91), with 7 of the 13 patients (54%) who were diagnosed with an SPM dying within 5 years. CSS for childhood neuroblastoma has improved substantially over time in Australia, but still remains lower than for most other types of childhood cancer. SPMs are uncommon and carry a better prognosis than relapse of the primary tumour.

Renal tumours in Australian children: 30 years of incidence, outcome and second primary malignancy data from the Australian Childhood Cancer Registry.

This paper describes the incidence and outcomes of childhood renal malignancies in Australia using national population-based data from the Australian Childhood Cancer Registry. De-identified data for children (0-14 years) diagnosed with renal malignancies from 1983 to 2015 inclusive were extracted. Cause-specific (CSS) and event-free survival up to 20 years from diagnosis were estimated using the cohort method. Adjusted excess mortality hazard ratios were calculated using a multivariable flexible parametric survival model. Details relating to second primary malignancies (SPMs) were also examined. There were 1046 children diagnosed with renal malignancies in Australia between 1983 and 2015 (91% nephroblastoma), generating an annual age-standardised incidence rate of 8 per million children, which remained constant over the study period. CSS was 89% (95% confidence interval = 87-91%) and 88% (86-90%) at 5 and 20 years, respectively, and 5-year event-free survival was 82% (80-84%). Five-year CSS did not change over the study period and was highest for nephroblastoma (91%). Of the 94% of patients achieving remission, 15% relapsed and subsequent 5-year CSS was 49% (40%-58%). Eleven children were diagnosed with SPM (standardised incidence ratio = 2.9, 95% confidence interval = 1.6-5.3, P < 0.001), and five of them (45%) died within 5 years of the second diagnosis. Children treated for renal malignancies in Australia have excellent long-term survival, which is unchanged since 1983. SPMs are uncommon following treatment for childhood renal cancer but carry a poor prognosis. Relapse carries a similarly poor prognosis to SPM but is more common. These data are comparable to registry outcomes in similarly developed nations.

A Pilot Study of Radiotherapy and Local Hyperthermia in Elderly Patients With Muscle-Invasive Bladder Cancers Unfit for Definitive Surgery or Chemoradiotherapy.

Purpose: To present the outcomes of a pilot study with hyperthermia (HT) and radiotherapy (RT) in elderly patients of muscle-invasive bladder cancers (MIBC) unfit for surgery or chemoradiotherapy (CTRT).Methods: Sixteen elderly patients with unifocal or multifocal MIBCs received a total dose of 48 Gy/16 fractions/4 weeks or 50 Gy/20 fractions/4 weeks, respectively. HT with a radiofrequency HT unit was delivered once weekly for 60 min before RT and a mean temperature of 41.3°C was attained (maximum temperature 41.1–43.5°C). Local control was assessed using RECIST criteria at 3-monthly intervals by cystoscopy with or without biopsy.Results: The median age, KPS and age-adjusted Charlson comorbidity index were 81 years, 70 and 5, respectively. At median follow-up of 18.5 months (range: 4–65), bladder preservation was 100% with satisfactory function. 11/16 patients (68.7%) had no local and/or distant failure, while isolated local, distant and combined local and distant failures were evident in 2, 2, and 1 patient, respectively. Two local failures were salvaged by TUR-BT resulting in a local control rate of 93.7%. The 5-year cause-specific (CS) local disease free survival (LDFS), disease free survival (DFS), and overall survival (OS) were 64.3, 51.6, and 67.5%, respectively while 5-year non-cause-specific (NCS)-LDFS, NCS-DFS, and NCS-OS were 26.5, 23.2, and 38%, respectively. None of the patients had acute or late grade 3/4 gastrointestinal or genitourinary toxicities.Conclusions: The outcomes from this pilot study indicate that thermoradiotherapy is a feasible therapeutic modality in elderly MIBC patients unfit for surgery or CTRT. HTRT is well-tolerated, allows bladder preservation and function, achieves long-term satisfactory locoregional control and is devoid of significant treatment-related morbidity. This therapeutic approach deserves further evaluation in randomized studies.

Underutilization of brachytherapy and disparities in survival for patients with cervical cancer in California

PurposeThe treatment for locally advanced cervical cancer is external beam radiation (EBRT), concurrent chemotherapy, and brachytherapy (BT). We investigated demographic and socioeconomic factors that influence trends in BT utilization and disparities in survival. MethodsUsing the California Cancer Registry, cervical cancer patients FIGO IB2-IVA from 2004 to 2014 were identified. We collected tumor, demographic and socioeconomic (SES) factors. We used multivariable logistic regression analysis to determine predictors of use of BT. Using Cox proportional hazards, we examined the impact of BT vs EBRT boost on cause specific (CSS) and overall survival (OS). ResultsWe identified 4783 patients with FIGO stage 11% IB2; 32% II, 54% III, 3% IVA. Nearly half (45%) of patients were treated with BT, 18% were treated with a EBRT boost, and 37% had no boost. Stage II and III were more likely to be treated with BT (p = 0.002 and p = 0.0168) vs Stage IB2. As patients aged, the use of BT decreased. Using multivariate analysis, BT impacted CCS (HR 1.16, p = 0.0330) and OS (HR 1.14, p = 0.0333). Worse CSS was observed for black patients (p = 0.0002), low SES (p = 0.0263), stage III and IVA (p < 0.0001. Black patients, low and middle SES had worse OS, (p = 0.0003). ConclusionsThe utilization of BT in locally advanced cervical cancer was low at 45%, with a decrease in CSS and OS. Black patients and those in low SES had worse CSS. As we strive for outcome improvement in cervical cancer, we need to target increasing access and disparities for quality and value.

Surgical therapy of thymic tumours with pleural involvement: an ESTS Thymic Working Group Project.

OBJECTIVES: Surgery for thymic epithelial tumours (TETs) with pleural involvement is infrequently performed. Thus, the value of surgical therapy for primary or recurrent TETs with pleural involvement is not sufficiently defined yet. METHODS: Twelve institutions contributed retrospective data on 152 patients undergoing surgery (1977–2014) on behalf of the ESTS Thymic Working group. Outcome measures included overall (OS), cause-specific (CSS) and disease-free (DFS) survival as well as freedom from recurrence (FFR). RESULTS: In 70.4% of cases, pleural involvement was present at the time of primary intervention, whereas 29.6% had surgery for recurrent disease involving the pleura. Pleural involvement resulted from thymomas (88.8%) and thymic carcinomas (11.2%). Forty extrapleural pneumonectomies (EPPs), 23 total pleurectomies (TPs), and 88 local pleurectomies (LPs) were performed (completeness of resection in 76.8%). OS for the entire patient population at 1, 3, 5 and 10 years was 96.4%, 91.0%, 87.2% and 62.7%, respectively. There was no statistically significant difference regarding FFR and OS for patients with local or advanced disease undergoing EPP, TP or LP. Thymic carcinomas in comparison with thymomas had a negative impact on OS [hazard ratio 6.506, P = 0.002], CSS and FFR. Incomplete resections predicted worse OS [hazard ratio 6.696, P = 0.003]. CONCLUSIONS: Complete resection remains the mainstay of treatment for TETs with pleural involvement. Study populations treated with EPP, TP and LP had similar survival that may be factual as observed, but in the presence of selection bias, we can further conclude from the results that EPP, TP and LP are equally effective procedures. Procedural choice depends upon the extent of tumour distribution. EPPs, TPs and LPs performed within a multimodality setting seem to be efficient procedures for local control of disease, as they yield excellent results regarding OS, DFS, CSS and FFR.

Modal Choice in the Pittsburgh Metropolitan Statistical Area: An Exploratory Data Analysis

Journey-to-work mode choice is intertwined with ideological and pragmatic issues. This article reexamines such issues using socioeconomic data from the decennial census and American Community Survey (ACS). It investigates the structure of variables with exploratory data analysis (EDA) because this technique advises the formation of hypotheses and the specification of cause and effect. Traditional EDA reveals the nonnormal structure of raw data, mapping illustrates associations between transit and income, and both methods suggest the presence of a transit-by-choice population among affluent metropolitan residents. The results yield three hypotheses concerning propensity to use transit that have previously received little attention.

Outcomes and prognostic factors for major salivary gland carcinoma following postoperative radiotherapy

To report outcomes of postoperative radiotherapy (PORT) for major salivary gland carcinoma (SGC) and identify patients at high risk of distant metastases (DM). Patients with major SGC treated between 2000-2012 were identified. All patients underwent initial primary resection, with neck dissection (ND) therapeutically (if N+) or electively in high risk N0 patients. PORT was delivered using 3D-CRT or IMRT. Multivariable analysis (MVA) assessed predictors for DM, cause-specific (CSS) and overall survival. Overall 304 patients were identified: 48% stage III-IVB, 22% lymphovascular invasion (LVI), 50% involved margins and 64% high risk pathology. ND was performed in 154 patients (51%). Adjuvant chemotherapy was used in 10 patients (3%). IMRT was delivered in 171 patients (56%) and 3D-CRT in 133 (44%). With a median follow-up of 82 months, the 5-(10-) year local, regional, distant control, CSS and OS were 96% (96%), 95% (94%), 80% (77%), 83% (82%) and 78% (75%), respectively. DM was the most frequent treatment failure (n=62). On MVA, stage III-IVB and LVI significantly correlated with DM, CSS and OS, while positive margins predicted DM and CSS, and high risk pathology predicted DM. No grade ⩾ 4 RTOG late toxicity was reported; 9 patients had grade 3, including osteoradionecrosis (n=4), neck fibrosis (n=3), trismus (n=1) and dysphagia (n=1). Surgery and PORT with 3D-CRT/IMRT produced excellent long-term outcomes. Further research is required for patients with stage III-IVB, LVI, positive margins and high risk pathology to determine the incremental benefit of systemic therapy in management of SGC.

Association between hypnotics use and increased mortality: causation or confounding?

Many research studies have found associations between benzodiazepines and/or z-hypnotics (BZZ) and increasing mortality, leading to a discussion about causation or confounding. This study suggests a factor that could produce this association through confounding. The Norwegian population in 2010 supplied 8862 deaths ages 41-80 and 898,289 controls. Index dates were added to control records which corresponded to death dates. BZZ use was recorded for 2 years before death/index date. Persons exposed to BZZ were more likely (OR = 2.3) to die than those who were not. With proximity of death, increasingly larger proportions of the prospective deaths received prescriptions for BZZ, until in the last 2 months 40-45% received BZZ. The frequency of BZZ use in controls increased with age as opposed to the death cohort where all ages showed similar rates of BZZ use. In the last few months before death, the youngest age group had an OR = 5.8 for BZZ use while the oldest age group an OR = 1.8, adjusted for age and sex. Opioid use showed a similar pattern of increasing use near death. The increased use of BZZ with approaching death is consistent with increasing symptomatic treatment in terminal illness. Thus, the association of BZZ and mortality is more likely to be due to confounding than to causality. Further evidence from this and other research includes similar use patterns for other drugs such as opioids, the lack of specificity in cause of death and the size of the association regarding age and time to death.

Clinical efficacy of accelerated partial-breast irradiation in treatment of ER-negative breast cancer: Results of a matched pair analysis.

70 Background: To assess outcomes of ER-negative breast CA pts treated with APBI, a matched-pair analysis was performed to determine efficacy of APBI vs whole breast RT (WBRT) from a single institution. Methods: From over 1650 pts treated with BCT from 1980-2013, a cohort of ER[-] pts treated with APBI or WBRT were investigated. Matched-pair analysis with a 1:1 ratio paired 79 APBI with 79 WBRT pts, all ER[-] (total:158). Match criteria included follow-up (FU) &gt; 1.0 yr, stage, &amp; age +/- 5 yrs. Outcomes analyzed included local recurrence (LR), true recurrence/marginal miss (TRMM), regional recurrence (RR), distant metastases (DM), disease-free (DFS), cause-specific (CSS), and overall survivals (OS). Results: As for clinical-pathological traits, no significant differences were noted for age (p=0.302), T-stage (p=1.000), tumor size (p=0.721), N-stage (p=0.062), use of chemoRx (p=0.747), endocrine Rx(p=0.408) or Herceptin (p=1.00). Per ASTRO Guidelines, no differences were seen in cautionary or unsuitable [UnS] groups between APBI &amp; WBRT (p=0.333). With a mean FU of 8.0 yrs (10.1 yrs APBI; 8.4 yrs WBRT p&lt;0.001), no differences were seen in the 10-yr actuarial rates of LR (9.3% vs 22.1% p=0.094), RR (1.3% vs 8.1% p=0.299), DM (7.1% vs 13.0% p=0.429), DFS (83.9% vs. 72.5% p=0.214), CSS (93.5% vs. 89.0 % p=0.677), or OS (79.6% vs. 80.1% p=0.573) between APBI &amp; WBRT. Only TRMM was significantly different (0% APBI vs 12.5% p=0.011). In stratifying patients based on ER% (0%, 1-3%, 4-8%) no outcome differences were noted. Of the 158 ER[-] pts, 124 were cautionary with similar 10-yr outcomes except for TRMM (0% APBI;WBRT 14.4% p=0.017) &amp; CLBF (0% APBI;WBRT17.1% p=0.019). For the 34 UnS patients, no endpoint differences were seen APBI vs WBRT. But, when the entire 158 ER[-] patients were analyzed for # of UnS factors, increasing UnS factors led to significant risk of RR (p&lt;0.001) &amp; DM (p=0.002). Conclusions: With 10-year FU of APBI for ER[-], the clinical results were equivalent to WBRT. No differences were noted based on ER%. Increasing number of unsuitable factors had more RR and DM. Maturation of randomized trial data will be needed to provide Class I evidence for equivalence of APBI to WBRT in ER[-] patients.

Abstract P5-14-14: Trimodality therapy for inflammatory breast cancer: The fox chase experience

Abstract Objective: To evaluate the outcomes and toxicity of inflammatory breast cancer (IBC) patients (pts) treated with trimodality therapy. Materials and Methods: A retrospective chart review of 71 non-metastatic women with IBC treated using once daily (QD) radiation fractionation between 1990 and 2011 was performed. Eighty-six% of pts underwent ER/PR receptor status testing, and 43% underwent HER-2 testing. All pts underwent a course of neoadjuvant chemotherapy. Forty-three pts had adjuvant hormonal therapy, and 7 had adjuvant trastuzumab. All pts underwent mastectomy with axillary lymph node dissection followed by chest wall irradiation to a median dose of 50 Gy (34-50), and 46 Gy to supraclavicular region in 2-Gy QD fractions. Internal mammary nodes were treated in 3 pts (4%) and a posterior axillary boost (PAB) was added in 32 (45%) cases. Scar boosts (SB) were employed in 6 patients. A univariate (UVA) analysis using Kaplan-Meier estimation method included cN, pN, pT stage, PAB, SB, dermal lymphatic invasion, lymphovascular invasion, body mass index, response to chemo (RC), close/positive surgical margins (SM), age, and triple negative (TN) receptor status. Significant predictors on UVA were included in multivariate (MVA) analysis using the Cox proportional hazard model. Primary endpoints included local-regional recurrence (LRR), distant metastasis (DM), overall survival (OS), cause specific (CSS) and disease-free survival (DFS). Acute and chronic skin reactions and lymphedema were evaluated. Results: Median follow-up was 34 months (2-265) and the median age was 56 years (36-82). Actuarial 3 and 5 year OS was 63% and 43%, and DFS was 51% and 43%, respectively. The median time to first failure, LRR, and DM was 20 mo (1.5-147), 13.5 mo (9-18), and 21 mo (1.5-147), respectively. The first site of failure was distant for 34 (48%), and local-regional for 4 pts (6%). Twenty-five pts (35%) were without failure at the time of last follow up. UVA analysis for LRR revealed SM (p = 0.05) and RC (p = 0.002) to be significant predictors of failure, however, these were not significant on MVA. UVA for DM identified SM (p = 0.03), TN (p&amp;lt;0.0001), and pN (p = 0.03) to be significant predictors, with TN (p = 0.002) being the only significant variable on MVA. Significant predictors for OS on UVA were TN (p = 0.04), pN1 vs pN2 disease (p = 0.02), SM (p = 0.0008) and RC (p = 0.008). None of these remained significant on MVA. Significant predictors for CSS on UVA were TN (p = 0.008), pN (p = 0.05), SM (p = 0.0004) and RC (p = 0.002). Only SM (p = 0.01) was predictive on MVA. Significant predictors for DFS on UVA were SM (p = 0.0003), pN1 vs pN2 (p = 0.05), RC (p = 0.0006) and TN (p = 0.0001). Only RC (p = 0.005) and TN (p = 0.001) remained significant on MVA. Twelve pts developed dry, and 14 developed moist desquamation. Three pts developed CTCAE v3 G3 lymphedema, and 5 pts experienced G3 fibrosis. Conclusions: We found that TN predicted for a higher rate of DM and worse DFS. Improved CR predicted for better DFS. Age &amp;lt; 45, SM, pN2 disease and poor RC were not predictive of LRR. Use of SB did not improve LRR, although this cohort was small. Overall, our results show encouraging 3 and 5 year OS rates and acceptable toxicity for IBC pts treated with trimodality therapy. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P5-14-14.

InterVA-4 as a public health tool for measuring HIV/AIDS mortality: a validation study from five African countries

BackgroundReliable population-based data on HIV infection and AIDS mortality in sub-Saharan Africa are scanty, even though that is the region where most of the world’s AIDS deaths occur. There is therefore a great need for reliable and valid public health tools for assessing AIDS mortality.ObjectiveThe aim of this article is to validate the InterVA-4 verbal autopsy (VA) interpretative model within African populations where HIV sero-status is recorded on a prospective basis, and examine the distribution of cause-specific mortality among HIV-positive and HIV-negative people.DesignData from six sites of the Alpha Network, including HIV sero-status and VA interviews, were pooled. VA data according to the 2012 WHO format were extracted, and processed using the InterVA-4 model into likely causes of death. The model was blinded to the sero-status data. Cases with known pre-mortem HIV infection status were used to determine the specificity with which InterVA-4 could attribute HIV/AIDS as a cause of death. Cause-specific mortality fractions by HIV infection status were calculated, and a person-time model was built to analyse adjusted cause-specific mortality rate ratios.ResultsThe InterVA-4 model identified HIV/AIDS-related deaths with a specificity of 90.1% (95% CI 88.7–91.4%). Overall sensitivity could not be calculated, because HIV-positive people die from a range of causes. In a person-time model including 1,739 deaths in 1,161,688 HIV-negative person-years observed and 2,890 deaths in 75,110 HIV-positive person-years observed, the mortality ratio HIV-positive:negative was 29.0 (95% CI 27.1–31.0), after adjustment for age, sex, and study site. Cause-specific HIV-positive:negative mortality ratios for acute respiratory infections, HIV/AIDS-related deaths, meningitis, tuberculosis, and malnutrition were higher than the all-cause ratio; all causes had HIV-positive:negative mortality ratios significantly higher than unity.ConclusionsThese results were generally consistent with relatively small post-mortem and hospital-based diagnosis studies in the literature. The high specificity in cause of death attribution achieved in relation to HIV status, and large differences between specific causes by HIV status, show that InterVA-4 is an effective and valid tool for assessing HIV-related mortality.

BRAFV600E Mutation Analysis in Papillary Thyroid Carcinoma: Is it Useful for all Patients?

The BRAF(V600E) mutation has been adopted as a prognostic factor in papillary thyroid carcinoma (PTC). However, it remains unclear whether routine BRAF mutation analysis is useful in establishing a prognosis for PTC patients. In the present study we investigated BRAF mutation analysis in a large number of PTC patients with long-term follow-up. We enrolled 766 patients from our hospital who underwent initial surgery for PTC without distant metastasis at diagnosis between 1996 and 2001, and whose BRAF mutation of primary lesions could be analyzed. The average age and follow-up period were 51years and 130months, respectively. To date, 77 (10%) and 37 (5%) patients have developed lymph node and distant recurrence, respectively, and 10 (1%) have died of PTC. The BRAF mutation was positive in 281 patients (37%), and it had no prognostic impact on lymph node recurrence-free (LNRFS) (p=0.700), distant recurrence-free (DRFS) (p=0.696), and cause-specific (CSS) (p=0.125) survival in our entire series. However, CSS of BRAF mutation-positive high-risk patients based on AMES (p=0.030), MACIS (score >6) (p=0.017), the UICC stage (IVa) (p=0.021), CIH classification (Sugitani et al. Surgery 135:139-148, 2004) (p=0.015), and our own classification system (Ito et al. World J Surg 34:2570-2580, 2010) (p=0.010) were significantly poorer than CSS of mutation-negative high-risk patients. The BFAF mutation did not affect CSS of non-high-risk patients, although the incidence of the BRAF mutation did not significantly differ between high-risk and non-high-risk groups based on these classification systems. The BRAF mutation was not related to LNRFS and DRFS in the subsets of high-risk and non-high-risk patients. BRAF mutation analysis is useful in estimating the CSS of high-risk PTC patients based on the representative classification systems. It was not related to the prognosis in non-high-risk patients, at least those living in Japan.

Intraoperative Electron Beam Radiotherapy (IOERT) in the management of locally advanced or recurrent cervical cancer

BackgroundTo report outcomes in women with locally recurrent or advanced cervical cancer who received intraoperative electron beam radiotherapy (IOERT) as a component of therapy.MethodsFrom 1983 to 2010, 86 patients with locally recurrent (n = 73, 85%) or primary advanced (n = 13, 15%) cervical cancer received IOERT following surgery. Common surgeries included pelvic exenteration (n = 26; 30%) or sidewall resection (n = 22; 26%). The median IOERT dose was 15 Gy (range, 6.25-25 Gy). Sixty-one patients (71%) received perioperative external beam radiotherapy (EBRT; median dose, 45 Gy). Forty-one patients (48%) received perioperative chemotherapy.ResultsMedian follow-up was 2.7 years (range, 0.1-25.5 years). Resections were classified as R0 (n = 35, 41%), R1 (n = 30, 35%), or R2 (n = 21, 24%). Cumulative incidences of central (within the IOERT field) and locoregional relapse at 3 years were 23 and 38%, respectively. The 3-year cumulative incidence of distant relapse was 43%. Median survival was 15 months, and 3-year Kaplan-Meier estimates of cause-specific (CSS) and overall survival (OS) were 31 and 25%, respectively. On multivariate analysis, pelvic exenteration (p = 0.02) and perioperative EBRT (p = 0.009) were associated with improved central control in patients with recurrent disease. Recurrence within 6 months of initial therapy was associated with reduced CSS (p = 0.001). Common IOERT-related toxicities included peripheral neuropathy (n = 16), ureteral stenosis (n = 4), and bowel fistula/perforation (n = 4). Eleven of 16 patients with neuropathy required long-term pain medication.ConclusionsLong-term survival is possible with combined modality therapy including IOERT for advanced cervical cancer. Distant relapse is common, yet a significant number of patients experienced local progression in spite of aggressive treatment. In addition to consideration of disease- and treatment-related morbidity, other factors to be considered when selecting patients for this approach include the time interval from initial therapy to recurrence and whether the patient is able to receive perioperative EBRT and pelvic exenteration in addition to IOERT.

1202 FACTORS AFFECTING 15-YEAR SURVIVAL FOLLOWING PROSTATE BRACHYTHERAPY

You have accessJournal of UrologyProstate Cancer: Localized1 Apr 20131202 FACTORS AFFECTING 15-YEAR SURVIVAL FOLLOWING PROSTATE BRACHYTHERAPY Nelson Stone and Richard Stock Nelson StoneNelson Stone New York, NY More articles by this author and Richard StockRichard Stock New York, NY More articles by this author View All Author Informationhttps://doi.org/10.1016/j.juro.2013.02.2556AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookTwitterLinked InEmail INTRODUCTION AND OBJECTIVES We analyzed the factors affecting 15-year cause specific and all cause survival in a cohort of men treated by prostate brachytherapy (PB). METHODS 1661 men with T1-T3 prostate cancer with a median age of 66 years (range 39-85) were treated with PB and followed a mean of 10 years (range 5-19). Patients were stratified into 3 risk groups based on NCCN criteria and were treated with implant alone (NCCN 1), implant plus either hormone therapy (HT) or external beam irradiation (EBRT) (NCCN 2) or implant plus HT plus EBRT (NCCN 3). HT was administered to 898 (54%) for a median of 6 months (range 1-34). Post-implant radiation dose was calculated using the biological effective dose (BED). Estimates for cause specific (CSS) and all cause survival (ACS) were determined by Kaplan Meier method with comparisons by logistic regression and Cox proportions hazard rates. RESULTS The 10 and 15-year survival rates were 98.1% and 94.1% (CSS) and 86.8% and 57% (ACS). 10 and 15-year CSS by NCCN risk groups were 99.8% and 96.3% (1); 97.7% and 97.5% (2), and 92.5% and 85.2% (3). Gleason score (GS), PSA and clinical stage (CS) were significant predictors of CSS on univariate analysis (p<0.001), while HT and BED were not. Cox regression revealed GS (p<0.001) and CS (p=0.002) as significant. At the time of this analysis 37 (2.2%) men had died of prostate cancer. Significant variables affecting ACS included HT (p=0.009), diabetes (0.016), atrial fibrillation (0.008) and heart disease (p=0.03). A history of smoking, CAD, alcohol use, asthma, other cancer, HBP, stokes and emphysema did not influence ACS. Cox regression revealed HT (p=0.007, HR 1.43), diabetes (p=0.021, HR 1.74), atrial fib (p=0.027, HR 3.09) and heart disease (p=0.053, HR 1.94) as significant. HT decreased ACS at 15 years from 60.3% to 54.9% (p=0.009). There was no decrease in ACS in patients given HT for < 6 months (p=0.140) vs. > 6 months (p=0.005). This difference persisted in younger (age < 66, p=0.017) and older men (p=0.05). 236 (14.2%) men had died at the time of this analysis. CONCLUSIONS PB yields favorable 15-year CSS, especially in high risk patients. ACS is less in patients with preexisting diabetes, atrial fibrillation and heart disease. The use of HT for longer than 6 months has a negative effect on ACS even in younger patients without an apparent beneficial effect on CSS. © 2013 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 189Issue 4SApril 2013Page: e492-e493 Advertisement Copyright & Permissions© 2013 by American Urological Association Education and Research, Inc.MetricsAuthor Information Nelson Stone New York, NY More articles by this author Richard Stock New York, NY More articles by this author Expand All Advertisement Advertisement PDF downloadLoading ...

Evolving trends in non-clear cell renal cell cancer (RCC) epidemiology: A large registry analysis of 4,483 patients.

4607 Background: Non-clear cell (ncc) RCC is uncommon but includes a heterogeneous group of histologic subtypes such as papillary, chromophobe (Chr), medullary (Med), and collecting duct (CD) cancers. We used a large cancer registry to define nccRCC clinical features and outcomes, identify prognostic variables, and to generate hypotheses for further study. Methods: Invasive RCC tumors in the California Cancer Registry from 1998-2009 among adults &gt; 18 years of age (n=38,251) were analyzed. Baseline clinical and tumor variables were collected. Primary outcome measures were 3-year cause-specific (CSS) and overall survival (OS). Uni- and multivariate survival analysis were used to identify predictors of CSS and OS. Results: Of 38,251 RCC cases, 19,149 (50%) were of clear cell type; 14,619 (38.2%) were “unclassified.” Thus, 4,483 (11.7%) nccRCC cases were identified and included in this analysis. Of these, 3,304 (73.7%) were diagnosed in 2004-09, suggesting a shift to more precise coding of histologic subtypes starting in the early 2000s. Histology distribution (n, %): papillary - 2,863 (63.9%); Chr - 1,507 (33.6%); and other including Med and CD - 113 (2.5%). Variables associated with significantly better OS and CSS (univariate analysis) were Chr histology, female sex, and higher socioeconomic status (SES). Significantly worse OS and CSS were seen in Med+CD, age &gt; 65 yrs, no nephrectomy (Nx), and higher stage. Non-hispanic blacks had significantly worse OS and CSS, while the “targeted therapy era” (2004-2009) was associated with better OS. Multivariate analysis showed the following to be independently associated with outcomes (all p &lt;0.001; Hazard Ratios [HR] shown for CSS and OS): Chr histology (0.48, 0.56), Med+CD histology (2.99, 2.42), no Nx (2.84, 3.18), regional stage (5.84, 1.98), distant stage (25.7, 7.67); and non-hispanic blacks (1.5,2 p=0.006; 1.25, p=0.03). Age &gt; 65 yrs (HR 1.78, p&lt;0.001) and high SES (HR 0.88, p=0.001) were associated with OS but not CSS. Conclusions: This is among the largest registry analyses of nccRCC ever performed, showing emerging trends in this uncommon RCC subset. Clinical variables associated with CSS and OS were identified that can inform the design of future clinical trials in nccRCC.

Highly-conformal intensity-modulated radiotherapy reduced toxicity without jeopardizing outcome in patients with paranasal sinus cancer treated by surgery and radiotherapy or (chemo)radiation

To report oncologic outcomes of patients with paranasal sinus cancer (PNSC) treated by surgery and radiotherapy or (chemo)radiation and to investigate the impact of improving the radiation technique on outcomes and toxicity. Between 1999 and 2010, 82 consecutive patients with PNSC were treated by surgery and radiotherapy or by definitive (chemo)radiation. Three-dimensional conformal (3DCRT) or highly-conformal intensity-modulated RT (IMRT) was used. Endpoints were local control (LC), regional control (RC), disease-free (DFS), cause-specific (CSS), and overall survival (OS), late toxicity, and quality-of-life (QoL). After median follow-up of 51 months, the 5-year actuarial rates of LC, RC, DFS, CSS, and OS were 74%, 94%, 56%, 64%, and 54%, respectively. Grade ≥ 2 late toxicity at 5-years was 28%. High T-stage and perineural invasion were significantly associated with poor LC and RT-technique with late toxicity. Late toxicity was significantly lowered using IMRT, compared to 3DCRT (17% vs. 52%, p<0.0001). Visual preservation were significantly improved using IMRT (88% and 65%, respectively, p=0.01). LC-rate was also improved by IMRT (80% vs. 64%, respectively, p=0.2). QoL-scores deteriorated during and shortly after treatment but returned in almost all scales to baseline levels within 3-12 months. Surgery and radiotherapy or definitive (chemo)radiation resulted in good LC rates but with high rate of late side-effects. However, late toxicity and permanent visual impairment were significantly decreased by using highly-conformal IMRT without jeopardizing outcome. The improvements in the therapeutic ratio achieved by using IMRT would allow dose escalation of RT to further improve outcomes.

Combined-modality treatment improved outcome in sinonasal undifferentiated carcinoma: single-institutional experience of 21 patients and review of the literature

The optimal treatment of sinonasal undifferentiated carcinoma (SNUC) remains unclear. We report our results on the outcome and toxicity of patients with SNUC treated by a combined modality and attempt to define the optimal treatment strategies by reviewing the literature. Between 1996 and 2010, 21 consecutive patients with SNUC were treated by any combination of surgery, chemotherapy and radiotherapy. End points were local control (LC), regional control (RC), disease-free (DFS), cause-specific (CSS) overall survival (OS), and late toxicity. Organ preservation was defined as visual preservation without orbital exenteration. After median follow-up of 54 months, the 5-year actuarial rates of LC, RC, DFS, CSS, and OS were 80, 90, 64, 74, and 74 % respectively. On multivariate analysis, T-stage and multimodality treatment approach correlated significantly with LC. Elective nodal irradiation was given to 42 % of high-risk node-negative patients. None of them developed regional failure. The overall 5-year incidence of grade ≥2 late toxicity was 30 %. Treatment-related blindness was significantly decreased in patients treated with intensity-modulated radiotherapy (IMRT), compared to 2D and 3D-conformal radiotherapy (3DCRT), with organ preservation rates of 86 and 14 % respectively (p = 0.006). We concluded that combined-modality treatment with three, or at least two, modalities resulted in good LC, but with high overall rate of late toxicity. However, the incidence of late toxicity and permanent visual impairment were decreased over time by the introduction of IMRT. Because of the improvement in therapeutic ratio achieved by using IMRT, this highly conformal radiation technique should be the standard of care in patients with SNUC.

Comparison of accelerated partial breast irradiation via multicatheter interstitial brachytherapy versus whole breast radiation

BackgroundBrachytherapy as adjuvant treatment for early-stage breast cancer has become widely available and offers patients an expedited treatment schedule. Given this, many women are electing to undergo brachytherapy in lieu of standard fractionation radiotherapy. We compare outcomes between patients treated with accelerated partial breast irradiation (APBI) via multicatheter interstitial brachytherapy versus patients who were also eligible for and offered APBI but who chose whole breast radiation (WBI).MethodsPatients treated from December 2002 through May 2007 were reviewed. Selection criteria included patients with pTis-T2N0 disease, ≤ 3 cm unifocal tumors, and negative margins who underwent breast conservation surgery. Local control (LC), cause-specific (CSS) and overall survival (OS) were analyzed.Results202 patients were identified in the APBI cohort and 94 patients in the WBI cohort. Median follow-up for both groups exceeded 60 months. LC was 97.0% for the APBI cohort and 96.2% for the WBI cohort at 5 years (ns). Classification by 2010 ASTRO APBI consensus statement categories did not predict worse outcomes.ConclusionAPBI via multicatheter interstitial brachytherapy provides similar local failure rates compared to WBI at 5 years for properly selected patients. Excellent results were seen despite the high fraction of younger patients (< 60 years old) and patients with DCIS.

Evolving trends in non-clear cell renal cell cancer (RCC) epidemiology: A large registry analysis of 4,483 patients.

374 Background: Non-clear cell (ncc) RCC is uncommon but includes a heterogeneous group of histologic subtypes such as papillary, chromophobe (Chr), medullary (Med), and collecting duct (CD) cancers. We used a large cancer registry to define nccRCC clinical features and outcomes, identify prognostic variables, and to generate hypotheses for further study. Methods: Invasive RCC tumors in the California Cancer Registry from 1998–2009 among adults &gt; 18 years of age (n=38,251) were analyzed. Baseline clinical and tumor variables were collected. Primary outcome measures were 3-year cause-specific (CSS) and overall survival (OS). Uni- and multivariate survival analysis were used to identify predictors of CSS and OS. Results: Of 38,251 RCC cases, 19,149 (50%) were of clear cell type; 14,619 (38.2%) were “unclassified”. Thus, 4,483 (11.7%) nccRCC cases were identified and included in this analysis. Of these, 3304 (73.7%) were diagnosed in 2004–09, suggesting a shift to more precise coding of histologic subtypes starting in the early 2000s. Histology distribution (n, %): papillary − 2,863 (63.9%); Chr − 1,507 (33.6%); and other including Med and CD − 113 (2.5%). Variables associated with significantly better OS and CSS (univariate analysis) were Chr histology, female sex, and higher socioeconomic status (SES). Significantly worse OS and CSS were seen in Med+CD, age &gt; 65 yrs, no nephrectomy (Nx), and higher stage. Non-hispanic blacks had significantly worse OS and CSS, while the “targeted therapy era” (2004–2009) was associated with better OS. Multivariate analysis showed the following to be independently associated with outcomes (all p &lt;0.001; Hazard Ratios [HR] shown for CSS and OS): Chr histology (0.48, 0.56), Med+CD histology (2.99, 2.42), no Nx (2.84, 3.18), regional stage (5.84, 1.98), distant stage (25.7, 7.67); and non-hispanic blacks (1.5,2 p=0.006; 1.25, p=0.03). Age &gt; 65 yrs (HR 1.78, p&lt;0.001) and high SES (HR 0.88, p=0.001) were associated with OS but not CSS. Conclusions: This is among the largest registry analyses of nccRCC ever performed, showing emerging trends in this uncommon RCC subset. Clinical variables associated with CSS and OS were identified that can inform the design of future clinical trials in nccRCC.