Royston-Parmar Models Research Articles

Intron 4\u20135 hTERT DNA Hypermethylation in Merkel Cell Carcinoma: Frequency, Association with Other Clinico-pathological Features and Prognostic Relevance

Merkel cell carcinoma (MCC) is an aggressive skin tumor with neuroendocrine differentiation, mainly affecting elderly population or immunocompromised individuals. As methylation of the human telomerase reverse transcriptase (mhTERT) has been shown to be a prognostic factor in different tumors, we investigated its role in MCC, in particular in intron 4–5 where rs10069690 has been mapped and recognized as a cancer susceptibility locus. DNA methylation analysis of hTERT gene was assessed retrospectively in a cohort of 69 MCC patients from the University of Bologna, University of Turin and University of Insubria. Overall mortality was evaluated with Kaplan-Meier curves and multivariable Royston-Parmar models. High levels of mhTERT (mhTERThigh) (HR = 2.500, p = 0.015) and p63 (HR = 2.659, p = 0.016) were the only two clinico-pathological features significantly associated with a higher overall mortality at the multivariate analysis. We did not find different levels of mhTERT between MCPyV (+) and (−) cases (21 vs 14, p = 0.554); furthermore, mhTERThigh was strongly associated with older age (80.5 vs 72 years, p = 0.026), no angioinvasion (40.7% vs 71.0%, p = 0.015), lower Ki67 (50 vs 70%, p = 0.005), and PD-L1 expressions in both tumor (0 vs 3%, p = 0.021) and immune cells (0 vs 10%, p = 0.002). mhTERT is a frequently involved epigenetic mechanism and a relevant prognostic factor in MCC. In addition, it belongs to the shared oncogenic pathways of MCC (MCPyV and UV-radiations) and it could be crucial, together with other epigenetic and genetic mechanisms as gene amplification, in determining the final levels of hTERT mRNA and telomerase activity in these patients.

PlGF (Placental Growth Factor) Testing in Clinical Practice: Evidence From a Canadian Tertiary Maternity Referral Center.

[Figure: see text].

Mediterranean Diet and Risk of Dementia and Alzheimer's Disease in the EPIC-Spain Dementia Cohort Study.

The Mediterranean diet (MD) has shown to reduce the occurrence of several chronic diseases. To evaluate its potential protective role on dementia incidence we studied 16,160 healthy participants from the European Prospective Investigation into Cancer and Nutrition (EPIC)-Spain Dementia Cohort study recruited between 1992–1996 and followed up for a mean (±SD) of 21.6 (±3.4) years. A total of 459 incident cases of dementia were ascertained through expert revision of medical records. Data on habitual diet was collected through a validated diet history method to assess adherence to the relative Mediterranean Diet (rMED) score. Hazard ratios (HR) of dementia by rMED levels (low, medium and high adherence levels: ≤6, 7–10 and ≥11 points, respectively) were estimated using multivariable Cox models, whereas time-dependent effects were evaluated using flexible parametric Royston-Parmar (RP) models. Results of the fully adjusted model showed that high versus low adherence to the categorical rMED score was associated with a 20% (HR = 0.80, 95%CI: 0.60–1.06) lower risk of dementia overall and HR of dementia was 8% (HR = 0.92, 0.85–0.99, p = 0.021) lower for each 2-point increment of the continuous rMED score. By sub-types, a favorable association was also found in women for non-AD (HR per 2-points = 0.74, 95%CI: 0.62–0.89), while not statistically significant in men for AD (HR per 2-points = 0.88, 0.76–1.01). The association was stronger in participants with lower education. In conclusion, in this large prospective cohort study MD was inversely associated with dementia incidence after accounting for major cardiovascular risk factors. The results differed by dementia sub-type, sex, and education but there was no significant evidence of effect modification.

Incidence of Dementia and Associated Factors in the EPIC-Spain Dementia Cohort.

Dementia has become a public health priority as the number of cases continues to grow worldwide. To assess dementia incidence and determinants in the EPIC-Spain Dementia Cohort. 25,015 participants (57% women) were recruited from three Spanish regions between 1992-1996 and followed-up for over 20 years. Incident cases were ascertained through individual revision of medical records of potential cases. Crude and age-adjusted incidence rates (IR) of dementia and sub-types (Alzheimer's disease (AD), and non-AD) were calculated by sex. Neelson-Aalen cumulative incidence estimates at 10, 15, and 20 years were obtained for each sex and age group. Multivariate Royston-Parmar models were used to assess independent determinants. Global IR were higher in women for dementia and AD, and similar by sex for non-AD. IR ranged from 0.09 cases of dementia (95% confidence interval: 0.06-0.13) and 0.05 (0.03-0.09) of AD per 1000 person-years (py) in participants below 60 years, to 23.2 (15.9-33.8) cases of dementia and 14.6 (9.1-33.5) of AD (per 1000 py) in those ≥85 years. Adjusted IR were consistently higher in women than men for overall dementia and AD. Up to 12.5% of women and 9.1% of men 60-65 years-old developed dementia within 20 years. Low education, diabetes, and hyperlipidemia were the main independent predictors of dementia risk, whereas alcohol showed an inverse association. Dementia incidence increased with age and was higher among women, but showed no geographical pattern. Dementia risk was higher among subjects with lower education, not drinking alcohol, and presenting cardiovascular risk factors.

Retrospective observational cohort study on innovation in oncology and progress in survival: How far have we gotten in the two decades of treating patients with advanced non-small cell lung cancer as a single population?

We assessed the impact of new antineoplastic agents on the overall survival (OS) of advanced non-small cell lung cancer (aNSCLC) patients followed up until 2012. Multivariate regression models were run for OS (outcome) and four proxies for innovation (exposure): Index (InnovInd, for SEER-Research data 1973–2012) and three levels of aggregation of Mean Medication Vintage, i.e. Overall (MMVOverall), using data aggregated at the State Level (MMVState), and using patient-level data (MMVPatient) using data from the US captured in SEER-Medicare 1991–2012. We derived Hazard ratios (HR) from Royston-Parmar models and odds ratios (OR) from a logistic regression on 1-year OS. Including 164,704 patients (median age 72 years, 56.8% stage IV, 61.8% with no comorbidities, 37.8% with adenocarcinoma, 22.9% with squamous-cell, 6.1% were censored). One-year OS improved from 0.22 in 1973 to 0.39 in 2012, in correlation with InnovInd (r = 0.97). Ten new NSCLC drugs were approved and 28 more used off-label. Regression-models results indicate that therapeutic innovation only marginally reduced the risk of dying (HROverall = 0.98 [0.98–0.98], HRMMV-Patient = 0.98 [0.97–0.98], and HRMMV-State = 0.98 [0.98–0.98], and slightly improved 1-year survival (ORMMV-Overall = 1.05 95%CI [1.04–1.05]). These results were validated with data from the Swedish National Health Data registers. Until 2013, aNSCLC patients were treated undifferentiated and the introduction of innovative therapies had statistically significant, albeit modest, effects on survival. Most treatments used off-guidelines highlight the high unmet need; however new advancements in treatment may further improve survival.

Parkinson's disease and patient related outcomes in stroke: A matched cohort study

ObjectiveTo evaluate post-stroke outcomes in patients with Parkinson's disease (PD). MethodsA matched cohort study was performed. Stroke patients with PD and non-PD controls were extracted from the Thailand Universal Insurance Database. Logistic regressions were used to evaluate the association between PD and in-hospital outcomes (mortality and complications). The PD-associated long-term mortality was evaluated using Royston-Parmar models. ResultsA total of 1967 patients with PD were identified between 2003 and 2015 and matched to controls (1:4) by age, sex, admission year, and stroke type. PD patients had decreased odds of in-hospital death: OR (95% CI) 0.66 (0.52 – 0.84) and 0.61 (0.43 – 0.85) after ischaemic and haemorrhagic strokes, respectively. PD was associated with a length-of-stay greater than median (4 days) after both stroke types: 1.37 (1.21 – 1.56) and 1.45 (1.05 – 2.00), respectively. Ischaemic stroke patients with PD also had increased odds of developing pneumonia, sepsis and AKI: 1.52 (1.2 – 1.83), 1.54 (1.16 – 2.05), and 1.33 (1.02 – 1.73). In haemorrhagic stroke patients, PD was associated with pneumonia: 1.89 (1.31 – 2.72). Survival analyses showed that PD was protective against death in the short term (HR=0.66; 95% CI 0.53–0.83 ischaemic, and HR=0.50; 95% CI 0.37 – 0.68 haemorrhagic stroke), but leads to an increased mortality risk approximately 1 and 3 months after ischaemic and haemorrhagic stroke, respectively. ConclusionPD is associated with a reduced mortality risk during the first 2–4 weeks post-admission but an increased risk thereafter, in addition to increased odds of in-hospital complications and prolonged hospitalisation.

Examining the Impact of First Nations Status on the Relationship Between Diabetes and Cancer.

Purpose: This population-based study examined the relationship between diabetes and cancer and determined if this relationship was influenced by First Nations (FN) status.Methods: In a matched case–cohort study, individuals 30–74 years of age diagnosed with diabetes during 1984–2008 in the province of Manitoba, Canada, with no cancer diagnosis before their diabetes diagnosis were matched to one diabetes-free control by age, sex, FN status, and residence. Flexible competing risk and Royston–Parmar regression models were used to compare cancer rates.Results: Overall, 72,715 individuals diagnosed with diabetes were matched to controls. In all age groups, diabetes was related to an increased risk of cancer. The relationship between diabetes and any type of cancer was not influenced by FN status (i.e., there was no interaction between the diagnosis of diabetes and people's FN status for any age group). The only significant interaction between diabetes and FN status was for kidney cancer for individuals 60–74 years of age; diabetes increased the risk of kidney cancer for all other Manitobans (AOMs) but not for FN.Conclusions: Diabetes increased the risk of cancer. The association was not modified by FN status except for kidney cancer where diabetes increased the risk for AOMs but not for FN.

Impact of Diabetes on Complications, Long Term Mortality and Recurrence in 608,890 Hospitalised Patients with Stroke.

Background:Patients with diabetes mellitus (DM) have been found to be at an increased risk of suffering a stroke. However, research on the impact of DM on stroke outcomes is limited.Objectives:We aimed to examine the influence of DM on outcomes in ischaemic (IS) and haemorrhagic stroke (HS) patients.Methods:We included 608,890 consecutive stroke patients from the Thailand national insurance registry. In-hospital mortality, sepsis, pneumonia, acute kidney injury (AKI), urinary tract infection (UTI) and cardiovascular events were evaluated using logistic regressions. Long-term analysis was performed on first-stroke patients with a determined pathology (n = 398,663) using Royston-Parmar models. Median follow-ups were 4.21 and 4.78 years for IS and HS, respectively. All analyses were stratified by stroke sub-type.Results:Mean age (SD) was 64.3 (13.7) years, 44.9% were female with 61% IS, 28% HS and 11% undetermined strokes. DM was associated with in-hospital death, pneumonia, sepsis, AKI and cardiovascular events (odds ratios ranging from 1.13–1.78, p < 0.01) in both stroke types. In IS, DM was associated with long-term mortality and recurrence throughout the follow-up: HRmax (99% CI) at t = 4108 days: 1.54 (1.27, 1.86) and HR (99% CI) = 1.27(1.23,1.32), respectively. In HS, HRmax (t = 4108 days) for long-term mortality was 2.10 (1.87, 2.37), significant after day 14 post-discharge. HRmax (t = 455) for long-term recurrence of HS was 1.29 (1.09, 1.53), significant after day 116 post-discharge.Conclusions:Regardless of stroke type, DM was associated with in-hospital death and complications, long-term mortality and stroke recurrence.

PD-1 (PDCD1) promoter methylation in Merkel cell carcinoma: prognostic relevance and relationship with clinico-pathological parameters

Merkel cell carcinoma is an aggressive neuroendocrine skin tumor, for which several non-conclusive prognostic factors of adverse clinical behavior have been reported. As promoter methylation of the immune checkpoint receptor CD279/PD-1/PDCD1(mPDCD1) has been shown to be a prognostic factor in different cancers, we investigated its role in Merkel cell carcinoma. mPDCD1was assessed retrospectively in a cohort of 69 Merkel cell carcinoma patients from the University of Bologna, University of Turin and University of Insubria. Kaplan-Meier curves and log-rank tests were calculated for all variables. To assess the influence of mPDCD1, the Cox proportional hazards model and different Royston-Parmar models were evaluated. High PDCD1 methylation (mPDCD1high) was associated with a higher overall mortality at both the univariate analysis (log rank test: χ2 = 5.17, p = 0.023; permutation test: p = 0.023) and the multivariate analysis (HR = 2.111, p = 0.042). The other variables associated with a higher overall mortality at the multivariate analysis were clinical stage III-IV (HR = 2.357, p = 0.008), size > 2 cm (HR = 2.248, p = 0.031) and Merkel cell polyomavirus (HR = 0.397, p = 0.015). Further, mPDCD1high was strongly associated with older age (81 vs 76 years, p = 0.042), absence of immune cells (92.6%, p < 0.001), no expression of PD-L1 by immune cells (70.4%, p = 0.041) and by both immune and tumor cells (70.4%, p = 0.001). mPDCD1 is a valid prognostic parameter in patients affected by Merkel cell carcinoma. In addition, it could provide an estimate of the global PD-1/PD-L1 expression with potentially relevant implications from a therapeutic point of view.

Chronic Disease Population Risk Tool (CDPoRT): a study protocol for a prediction model that assesses population-based chronic disease incidence

BackgroundPopulation-based risk prediction tools exist for individual chronic diseases. From a population health perspective, studying chronic diseases together provides a comprehensive view of the burden of disease in the population. Thus, public health officials and health policymakers would benefit from a prediction tool that measures the incidence of chronic diseases compositely. This study protocol proposes the development and validation of the Chronic Disease Population Risk Tool (CDPoRT) that will predict the incidence of six chronic diseases in the population setting using multivariable modeling techniques.MethodsCDPoRT will be built using population-based responses to the first six cycles of the Canadian Community Health Survey linked to health administrative data in Ontario and Manitoba from 2000 to 2014. Predictors including modifiable lifestyle risk factors (i.e., alcohol consumption, cigarette smoking, diet, and physical activity) will be used to predict time-to-chronic disease incidence (i.e., congestive heart failure, chronic obstructive pulmonary disease, diabetes, lung cancer, myocardial infarction, and stroke including transient ischemic heart attack). Sex-specific Royston-Parmar models will be used for model development and validation with death free of chronic disease as a competing risk. CDPoRT will be developed using an Ontario derivation cohort consisting of 47,960 females and 38,267 males with 7035 and 6220 chronic disease events, respectively. The model will be validated using split-sample validation using an Ontario validation cohort consisting of 20,325 females and 16,627 males with 2972 and 2658 chronic disease events, respectively. The model will be externally validated in the Manitoba validation cohort (i.e., geographic validation) expected to consist of 11,800 females and 9700 males with 1650 and 1550 chronic disease events, respectively. Measures of overall predictive accuracy (e.g., Nagelkerke’s R2), discrimination (e.g., Harrell’s concordance statistic), and calibration (e.g., calibration plots) will be used to assess predictive performance.DiscussionTo the extent of our knowledge, CDPoRT will be the first population-based regression prediction model that will predict the incidence of multiple chronic diseases simultaneously at the population level.

Bayesian one-step IPD network meta-analysis of time-to-event data using Royston-Parmar models.

Network meta‐analysis (NMA) combines direct and indirect evidence from trials to calculate and rank treatment estimates. While modelling approaches for continuous and binary outcomes are relatively well developed, less work has been done with time‐to‐event outcomes. Such outcomes are usually analysed using Cox proportional hazard (PH) models. However, in oncology with longer follow‐up time, and time‐dependent effects of targeted treatments, this may no longer be appropriate. Network meta‐analysis conducted in the Bayesian setting has been increasing in popularity. However, fitting the Cox model is computationally intensive, making it unsuitable for many datasets. Royston‐Parmar models are a flexible alternative that can accommodate time‐dependent effects. Motivated by individual participant data (IPD) from 37 cervical cancer trials (5922 women) comparing surgery, radiotherapy, and chemotherapy, this paper develops an IPD Royston‐Parmar Bayesian NMA model for overall survival. We give WinBUGS code for the model. We show how including a treatment‐ln(time) interaction can be used to conduct a global test for PH, illustrate how to test for consistency of direct and indirect evidence, and assess within‐design heterogeneity. Our approach provides a computationally practical, flexible Bayesian approach to NMA of IPD survival data, which readily extends to include additional complexities, such as non‐PH, increasingly found in oncology trials.

Parametric and penalized generalized survival models

We describe generalized survival models, where g( S( t| z)), for link function g, survival S, time t, and covariates z, is modeled by a linear predictor in terms of covariate effects and smooth time effects. These models include proportional hazards and proportional odds models, and extend the parametric Royston-Parmar models. Estimation is described for both fully parametric linear predictors and combinations of penalized smoothers and parametric effects. The penalized smoothing parameters can be selected automatically using several information criteria. The link function may be selected based on prior assumptions or using an information criterion. We have implemented the models in R. All of the penalized smoothers from the mgcv package are available for smooth time effects and smooth covariate effects. The generalized survival models perform well in a simulation study, compared with some existing models. The estimation of smooth covariate effects and smooth time-dependent hazard or odds ratios is simplified, compared with many non-parametric models. Applying these models to three cancer survival datasets, we find that the proportional odds model is better than the proportional hazards model for two of the datasets.

Do breast cancer survivors benefit from prophylactic removal of uterus and ovaries? A population-based data linkage replication study.

Our previous population-based research found prophylactic surgery (hysterectomy and bilateral salpingo-oophorectomy [BSO]) halved the mortality risk for premenopausal breast cancer patients. Here we aim to replicate findings in a Western Australia dataset. Data from the Western Australia Cancer Registry of 15395 women 20-79 years diagnosed with primary breast cancer (1997-2011) was categorized into four groups: neither hysterectomy nor BSO, hysterectomy only, BSO only, or hysterectomy + BSO. We fitted flexible parametric breast cancer-specific and overall survival models with 95% confidence intervals (also known as Royston-Parmar models) to assess the impact of prophylactic surgery. A total of 12630 (82.0%) patients had no surgery, 1799 (11.7%) had a hysterectomy only, 337 (2.2%) had BSO only and 629 (4.1%) had both a hysterectomy and BSO. For all-causes mortality, unadjusted 10-year survival was highest for women who had either a hysterectomy + BSO (84.7%) or a hysterectomy only (84.2%). After adjusting for covariates, the survival advantage compared to women without any surgery remained significant for the hysterectomy only group (hazard ratio [HR] = 0.89; 95% confidence interval [CI], 0.81-0.98; P = 0.02). A similar pattern emerged in breast cancer-specific survival with significantly improved survival for women who had a hysterectomy only (HR = 0.83; 95% CI, 0.74-0.94; P = 0.003). However, for non-breast cancer-related survival, having a BSO alone increased risk of death (HR = 1.83; 95% CI, 1.14-2.93; P = 0.01). We observed significantly improved overall and breast cancer-specific survival among women who had a hysterectomy only, but increased non-breast cancer-related risk after BSO only. Breast cancer patients must weigh up pros and cons of prophylactic surgery.

One-step individual patient data (IPD) network meta-analysis of survival data using royston-parmar models

Background Network meta-analysis (NMA) combines direct and indirect evidence from trials to calculate treatment effect estimates and rank treatments. Modelling approaches for continuous and binary outcomes are relatively well developed, but less work has been done with time-toevent outcomes. Such outcomes have usually been analysed using Cox proportional hazard (PH) models, but in oncology with longer follow-up of trials, and timedependent effects of targeted treatments, this may no longer be appropriate. Royston-Parmar models, fitted using WinBUGS, are a flexible alternative.

The impact of risk‐reducing hysterectomy and bilateral salpingo‐oophorectomy on survival in patients with a history of breast cancer—A population‐based data linkage study

Prophylactic surgery including hysterectomy and bilateral salpingo-oophorectomy (BSO) is recommended in breast cancer susceptibility gene (BRCA)-positive women, whereas in women from the general population, hysterectomy plus BSO may increase the risk of overall mortality. The effect of hysterectomy plus BSO on women previously diagnosed with breast cancer is unknown. We used data from a population-base data linkage study of all women diagnosed with primary breast cancer in Queensland, Australia between 1997 and 2008 (n = 21,067). We fitted flexible parametric breast cancer-specific and overall survival models with 95% confidence intervals (also known as Royston-Parmar models) to assess the impact of risk-reducing surgery (removal of uterus, one or both ovaries). We also stratified analyses by age 20-49 and 50-79 years, respectively. Overall, 1,426 women (7%) underwent risk-reducing surgery (13% of premenopausal women and 3% of postmenopausal women). No women who had risk-reducing surgery compared to 171 who did not have risk-reducing surgery developed a gynaecological cancer. Overall, 3,165 (15%) women died, including 2,195 (10%) from breast cancer. Hysterectomy plus BSO was associated with significantly reduced risk of death overall [adjusted hazard ration (HR), 0.69; 95% confidence interval (CI), 0.53-0.89; p = 0.005]. Risk reduction was greater among premenopausal women, whose risk of death halved (HR, 0.45; 95% CI, 0.25-0.79; p < 0.006). This was largely driven by reduction in breast cancer-specific mortality (HR, 0.43; 95% CI, 0.24-0.79; p < 0.006). This population-based study found that risk-reducing surgery halved the mortality risk for premenopausal breast cancer patients. Replication of our results in independent cohorts and subsequently randomised trials are needed to confirm these findings.

Survival Analysis of Irish Amyotrophic Lateral Sclerosis Patients Diagnosed from 1995–2010

IntroductionThe Irish ALS register is a valuable resource for examining survival factors in Irish ALS patients. Cox regression has become the default tool for survival analysis, but recently new classes of flexible parametric survival analysis tools known as Royston-Parmar models have become available.MethodsWe employed Cox proportional hazards and Royston-Parmar flexible parametric modeling to examine factors affecting survival in Irish ALS patients. We further examined the effect of choice of timescale on Cox models and the proportional hazards assumption, and extended both Cox and Royston-Parmar models with time varying components.ResultsOn comparison of models we chose a Royston-Parmar proportional hazards model without time varying covariates as the best fit. Using this model we confirmed the association of known survival markers in ALS including age at diagnosis (Hazard Ratio (HR) 1.34 per 10 year increase; 95% CI 1.26–1.42), diagnostic delay (HR 0.96 per 12 weeks delay; 95% CI 0.94–0.97), Definite ALS (HR 1.47 95% CI 1.17–1.84), bulbar onset disease (HR 1.58 95% CI 1.33–1.87), riluzole use (HR 0.72 95% CI 0.61–0.85) and attendance at an ALS clinic (HR 0.74 95% CI 0.64–0.86).DiscussionOur analysis explored the strengths and weaknesses of Cox proportional hazard and Royston-Parmar flexible parametric methods. By including time varying components we were able to gain deeper understanding of the dataset. Variation in survival between time periods appears to be due to missing data in the first time period. The use of age as timescale to account for confounding by age resolved breaches of the proportional hazards assumption, but in doing so may have obscured deficiencies in the data. Our study demonstrates the need to test for, and fully explore, breaches of the Cox proportional hazards assumption. Royston-Parmar flexible parametric modeling proved a powerful method for achieving this.

Radiotherapy in Stage I and II Extranodal Marginal Zone B-Cell Lymphoma of Mucosa-Associated Lymphoid Tissue (MALT): Analysis of the Surveillance, Epidemiology, and End Results (SEER) Database

AbstractAbstract 2053 Background:External beam radiotherapy (RT) is the treatment of choice for many patients with localized (stage IE-IIE) MALT lymphoma, primarily based on single-institution case series. There are no specific factors favoring surgical management or addition of systemic therapy in these cases. The purpose of this study was to evaluate the association of RT used as the initial treatment strategy with survival outcomes in a population-based cohort. Methods:We analyzed 6,111 cases of st I/IIE MALT lymphoma of the most common anatomical locations included in the SEER database, diagnosed between 1998 and 2009. Cases identified through autopsy or death certificate only, as well as those with unknown stage or RT status were excluded. RT was recorded only if administered as part of the initial treatment strategy within 12 months of diagnosis. 5-year relative survival was summarized for treated and untreated patients. For multivariate regression, flexible parametric Royston-Parmar models for overall survival were fitted for each site, adjusting for clinical and socioeconomic confounders. Age and surgical excision of the primary site were always included as variables. No records of progression-free survival, alternative or subsequent systemic treatments or watchful waiting strategy were available. Results:There were significant differences in the rates of RT utilization, baseline clinical characteristics and survival outcomes. Median age varied from 57 years (cutaneous lymphoma) to 69 years (breast). In multivariate models, RT was associated with improved survival in patients with ocular, cutaneous or gastric localization with no significant association for the pulmonary, intestinal, salivary, mammary and thyroid sites. The hazard ratio in gastric MALT was non-proportional and increased from 0.45 at 6 months to 0.97 at 24 months form diagnosis. Long-term prognosis for ocular, salivary and cutaneous thyroid lymphomas was excellent, but was less favorable for the gastrointestinal or pulmonary locations. Conclusions:The survival benefits of RT in early-stage MALT lymphoma are most evident for ocular and cutaneous sites of origin, and this modality can be advocated for treatment. The relative survival of 99–100% in those anatomical subtypes may indicate curative potential in patients who receive RT. In gastric MALT lymphoma the benefit is limited to the early period of two years of follow up. Inferior survival despite local radiotherapy in pulmonary and intestinal locations may suggest higher rates of occult disseminated disease and potential value of systemic therapy.Siten% receiving RT5-year relative survival (95% CI)HR for RT (95% CI)PAll patientsNo RTRTGastric249927%0.88 (0.86–0.91)0.87 (0.84–0.89)0.92 (0.86–0.95)0.45–0.97a0.0002Ocular101367%0.94 (0.89–0.97)0.84 (0.75–0.90)0.99 (0.46–1.00)0.56 (0.4–0.77)0.0004Salivary61544%0.92 (0.86–0.96)0.87 (0.79–0.92)0.98 (0.47–1.00)0.73 (0.49–1.11)0.14Intestine60011%0.88 (0.83–0.92)0.88 (0.82–0.92)0.87 (0.64–0.96)0.79 (0.42–1.46)0.45Skin51451%0.98 (0.86–1.00)0.90 (0.76–0.96)1.00 (0.97–1.00)0.47 (0.27–0.81)0.007Lung4969%0.87 (0.80–0.92)0.87 (0.80–0.92)0.92 (0.46–0.99)b0.66 (0.32–1.38)0.28Breast21041%0.90 (0.80–0.95)0.86 (0.73–0.93)0.95 (0.66–0.99)0.62 (0.31–1.22)0.17Thyroid16449%0.93 (0.82–0.97)0.90 (0.74–0.96)0.95 (0.64–0.99)0.6 (0.25–1.41)0.24anon-proportional hazard.bnot age-standardized due to small number of patients. Disclosures:No relevant conflicts of interest to declare.

Modeling Time in Medical Education Research: The Potential of New Flexible Parametric Methods of Survival Analysis

Time – the duration of a certain process or the timing of a specified event – plays a central role in many situations in medical research. Waiting time analysis (“survival analysis”) is a field of statistics providing the tools for solving the unique problems of such studies. In particular, waiting time analysis correctly handles the typical positively skewed distributions of waiting times as well as censored observations on study subjects for whom the target event does not occur before data collection ends. For decades, non-parametric Kaplan-Meier analysis and semiparametric Cox regression despite some inherent limitations have dominated waiting time analysis in medical contexts, while parametric models, although in principle offering important theoretical advantages, were scarcely applied in practice because of lacking flexibility. Recently, however, new flexible parametric methods (Royston-Parmar models) became available offering exciting new research potential. Surprisingly, although medical education research deals with a range of typical problems suited for waiting time analysis, the methods were rarely used in the past. By re-analyzing data from a previous investigation on study dropout of medical students, this is the first study demonstrating the usefulness and practical applications of waiting time analysis with special emphasis on Royston-Parmar models in a medical education research environment.

External Validation of Prognostic Models in Terminally Ill Patients,

Abstract 4186 Background:Over one million Medicare beneficiaries receive hospice care annually. However, besides the well documented advantages of hospice, many Americans do not enjoy maximum benefit from the hospice care. The fundamental reason for this is related to the inappropriate and poorly timed referral of terminally ill patients to hospice. As a result, many patients die within a few days of referral, while some live many years after the referral was made. Improvement in the accuracy of prognosis translates into superior quality of care. Predictions based on statistical modeling have been shown to be superior to physicians' prognostication. However, very few of these statistical models have been externally validated in terminally ill patients. Here we report the external validation of 5 most commonly used prognostication models in a cohort of terminally ill patients: 1) declining exponential approximation of life expectancy (DEALE) 2) study to understand prognoses and preferences for outcomes and risks of treatments (SUPPORT), 3) adjusted palliative performance scale (PPS), 4) adjusted Karnofsky performance scale index (Karnofsky) and 5) adjusted eastern cooperative oncology group performance status (ECOG). Methods:We retrospectively extracted data from 590 deceased patients enrolled in Tampa Bay Lifepath Hospice and Palliative Care starting January 2009 and going backwards to validate the prognostic models. Two research assistants extracted all data necessary to populate the model variables and two faculty members randomly checked 25% of the data for accuracy. The models were tested against observed survival duration. PPS, Karnofsky and ECOG risk scores were predicted using a flexible family of Royston-Parmar parametric models and adjusted for age, gender and presence of cancer. We utilized several metrics to assess the performance of these models. Specifically, we used the Brier score and scaled Brier score (which is very similar to the Pearson correlation coefficient R2), the area under the receiver operating characteristic curve (AUROC), and the Hosmer-Lemshow goodness-of-fit p-value (HL). Results:Brier scores were consistently below the non-informative level of 0.25 and AUROC significantly higher than the non-informative level of 0.5 for the adjusted PPS, Karnofsky and ECOG models (table 1). The HL p-value was consistently greater than 0.1 only for PPS. SUPPORT and DEALE models did not predict fit our data well for survival at day one and month one, two and six. The AUROC takes a value close to 0.5, even though the Brier scores were relatively low and HL p-value greater than 0.05, this value is significantly close to 0.5 for SUPPORT and DEALE models (table 1). [Display omitted] Conclusion:None of the prognostication models accurately predicated survival among our cohort of terminally ill patients. However, PPS consistently performed best in predicting survival in terminally ill patients followed by Karnofsky and ECOG. Disclosures:No relevant conflicts of interest to declare.