Presence Of Long-term Survivors Research Articles

Clinical characterisation and long-term survival of paediatric and juvenile lymphoma in cats: 33 cases (2008-2022).

The aims of this study were to describe the clinical presentation, tumour characteristics, responses to chemotherapy protocols and toxicity in a cohort of cats with lymphoma up to 18 months of age. In addition, the probability of long-term (>2 years) survival was explored. The medical records of client-owned cats aged up to 18 months diagnosed with lymphoma between 2008 and 2022 at five UK-based veterinary referral hospitals were reviewed. Thirty-three cats were included. The most common anatomical forms were mediastinal (42%), disseminated disease (30%) and renal (15%), with all cats having intermediate to large cell lymphoma. Three out of 29 cats tested were positive for FeLV but none for FIV. Twenty-six cats were treated with multi-agent chemotherapy protocols with complete and partial responses seen in 46% and 50% of cats, respectively. For this group, median progression-free survival was 133 days (95% confidence interval [Cl] 67 to 199) and median survival time was 268 days (95% Cl 106 to 430). Complete response to chemotherapy was associated with a longer progression-free survival. Seven cats were considered long-term survivors (>2 years). Chemotherapy was generally well tolerated with none of the long-term survivors suffering from chronic sequelae from cytotoxic treatment. Paediatric and juvenile cats with lymphoma showed a high response rate to multi-agent chemotherapy protocols with rare significant toxicities. The presence of long-term survivors may suggest a more favourable outcome in a subset of patients.

A Metaheuristic Approach to Parameter Estimation of a Flexible Parametric Mixture Cure Rate Model with Interval-Censored Data

A flexible parametric mixture cure model, called bi-lognormal cure rate model or simply BLN model, is defined and studied. The BLN model can be effectively used to analyze survival dataset in the presence of long-term survivors, especially when the dataset presents the underlying phenomenon of latent competing risks or when there is evidence that a bimodal hazard function is appropriated to described it, which are advantages over other cure rate models found in the literature. We discuss the maximum likelihood estimation for the model parameters considering interval-censored data through the differential evolution algorithm that is a nature-inspired computing metaheuristic used for global optimization of functions defined in multidimensional spaces. This approach is also used because the likelihood function of the model is multimodal and the direct application of gradient methods in this case is not ideal, since such methods are local search methods with a high chance of getting stuck at a local maximum when the starting point is chosen outside the basin of attraction of a global maximum. In addition, a simulation study was implemented to compare the performance of differential evolution algorithm with the performance of the Newton-Raphson algorithm in terms of bias, root mean square error, and the coverage probability of the asymptotic confidence intervals for the parameters. Finally, an application of the BLN model to real data is presented to illustrate that it can provide a better fit than other mixture cure rate models.

Order restricted inference for a multiple step‐stress model with long‐term survivors for a general family of distributions

AbstractIn this article, we propose a failure rate based step‐stress accelerated life testing (SSALT) model assuming that the time‐to‐event distribution belongs to a fairly general family of distributions and the underlying population consists of long term survivors. With increase in stress levels, it is expected that the mean time to the event of interest gets shortened leading to an order restriction among the mean times‐to‐event. We propose here a method of obtaining order restricted maximum likelihood estimators (MLEs) based on expectation maximization (EM) algorithm coupled with the method of generalized isotonic regression technique. Additionally, we address the testing of hypothesis problem for the presence of long term survivors in the underlying population based on both asymptotic and non‐asymptotic approaches. To illustrate the effectiveness of the proposed method, extensive simulation experiments are carried out and a real data set is analyzed.

Long-term survival projections of loncastuximab tesirine-treated patients in relapsed or refractory (R/R) diffuse large B-cell lymphoma (DLBCL).

e19551 Background: Loncastuximab tesirine (loncastuximab tesirine-lpyl; Lonca) is an FDA approved CD19-directed antibody-drug conjugate for R/R DLBCL. From the LOTIS-2 trial primary data cut (April 6, 2020), overall response rate was 48.3% and median overall survival (OS) was 9.9 months. The OS Kaplan-Meier (KM) plot displayed a survival plateau suggesting the presence of long-term survivors (LTS). Survival analyses were conducted on a more mature data cut (March 1, 2021; median follow-up = 1.7 years, follow-up completeness at median = 81%) to estimate the percentage of LTS and expected lifetime survival (mean OS) for lonca-treated patients. Methods: Consistent with studies of other R/R DLBCL treatments, identified through a literature review, parametric and mixture cure models were fit utilizing multiple distributions. Flexible cubic spline (hazard scale 1-3 knots) and non-mixture cure analyses were also conducted. Age- and gender-matched United States life table hazards were used in projections for LTS and to ensure modeled hazards were not less than the general population. Best-fit models were determined through fit statistics, KM and fitted curve overlays, and clinical plausibility. The best-fit model from each method was a candidate for overall best fit. A hybrid model following the best-fit parametric/spline model to a defined time point and switching to general population mortality was also constructed. Results: Mixture and non-mixture cure models fit best (individual best fits gamma and Weibull, respectively). Parametric and spline models (individual best fits log-normal and 2 knot models, respectively) did not fit the observed data well nor fit the clinical expectation of long-term survival. Due to better fit, spline models were used in the hybrid model. LTS from the mixture cure and non-mixture cure models were 24-26%. Mixture cure, non-mixture cure, and hybrid model with a 2-year switch point were consistent in survival predictions (6.11-6.23 years). In a sensitivity analysis with 3-year switch point in the hybrid model, the estimated survival was shorter due to the switch point being below the observed survival plateau. Table presents full survival results and fit statistics. Conclusions: The observed survival plateau suggests lonca-treated patients may include LTS. Mixture cure, non-mixture cure, and hybrid models fit the trial data well and align on survival projections (6.11-6.23 years). Additional follow-up may help refine the switch point of the hybrid model and confirm presence of LTS.[Table: see text]

Survival analysis of bank loans in the presence of long-term survivors

In this paper we model competing risks, default and early settlement events, in the presence of long term survivors and compare survival and logistic methodologies. Cause specific Cox regression models were fitted and adjustments were made to accommodate a proportion of long term survivors. Methodologies were compared using ROC curves and area under the curves. The results show that survival methods perform better than logistic regression methods when modelling lifetime data in the presence of competing risks and in the presence of long term survivors. Keywords: Cause specific, Competing risk, Consumer credit, Proportional hazards

P40.10 Brain Metastases in Patients With Non-Small Cell Lung Cancer Treated With Immunotherapy. Real World Data From a Tertiary Hospital in Spain

The metastatic involvement of the central nervous system (CNS) has been traditionally related with an ominous prognosis in non-small cell lung cancer (NSCLC). Immunotherapy (IO) has (dramatically changed) revolutionized the treatment of non-oncogene addicted NSCLC but its tumour effectivity in the CNS has not been well established. The aim of this study is to identify if CNS metastases in lung cancer patients continue to be a major problem and a major cause of mortality in patients treated with IO. We performed a retrospective, observational and comparative study. We collected information regarding clinical factors, treatment, CNS involvement and outcomes from 185 patients with metastatic NSCLC treated with immunotherapy in Hospital Universitario Puerta de Hierro Majadahonda between March, 2014 and March, 2020. We performed a statistical analysis in order to exclude significant differences between both groups. We compared outcomes from patients with and without brain metastasis. 22.2% of patients had brain metastasis (CNS involvement) at diagnosis or during the course of the disease. No significant differences were found between groups regarding sex, age, histology, PD-L1, IO line, type of IO and performance status (PS). No significant differences in survival (p = 0.437) nor in the presence of long-term survivors (p = 0.723 for 2 years or more) were found. Comparative data from both groups are shown in Table-1. Specific data from patients with CNS metastases are in Table-2.Table-1Comparative dataWithout CNS metastases (N=144)With CNS metastases (N=41)PSexM: 45 (31.3%) F: 99 (68.8%)M: 19 (46.3%) F: 22 (53.7%)0.073Age (median, years; range)64.0 (40.0-82.0)58.0 (44.0-78.0)0.114HistologyAdenocarcinoma: 87 (60.4%)Squamous cell carcinoma: 42 (29.2%)Carcinoma NOS: 15 (10.4%)Adenocarcinoma: 31 (75.6%)Squamous cell carcinoma: 7 (17.1%)Carcinoma NOS: 3 (7.3%)0.199PD-L1 (median)60.055.00.638ECOG PS0-1: 134 (93.1%) 2 or more: 6 (4.1%) Unknown: 4 (2.7%)0-1: 37 (90.2%) 2 or more: 2 (4.9%) Unknown: 2 (4.9%)0.170IO line1: 41 (28.5%) 2: 50 (34.7%) Subsequent: 53 (36.8%)1: 9 (22.0%) 2: 16 (39.0%) Subsequent: 16 (9.0%)0.553Type of IOImmunotherapy: 122 (84.7%)Chemoimmunotherapy: 15 (14.9%) IO + TKI: 7 (4.9%)Immunotherapy: 35 (85.4%)Chemoimmunotherapy: 5 (12.2%) IO + TKI: 1 (2.4%)0.768Overall survival, months (median, CI 95%)Global: 16.0 (7.7-24.3) 1 line: 18.0 (7.5-28.5) 2 line: 21.0 (7.0-35.0) Subsequent lines: 8.0 (4.3-11.7)Global: 15.0 (10.3-19.7) 1 line: 19.0 (11.7-26.3) 2 line: 20.0 (0.0-43.7) Subsequent lines: 6.0 (0.1-11.9)0.437 0.860 0.342 0.706Long-term survivors2 years or more: 28 (19.4%) 3 years or more: 8 (5.6%)5 years or more: 4 (2.8%)2 years or more: 9 (22.0%) 3 years or more: 4 (9.8%)5 years or more: 2 (4.9%)0.723 0.335 0.503 Open table in a new tab Table-2Specific data from patients with brain metastasesBrain metastases before starting IO treatment20 (50.0%)Number of radiological metastases1-2: 17 (41.5%)3-10: 9 (22.0%) More than 10: 12 (29.3%) Unknown: 3 (7.30%)Meningeal carcinomatosis3 (7.5%)Local treatmentsSurgery: 3 (7.5%) Whole brain radiotherapy (WBRT): 24 (60.0%) Radiosurgery: 13 (32.5%)Radiation necrosis1 (2.5%)Best CNS responseProgressive disease: 13 (31.7%) Stable disease: 5 (12.2%) Partial response: 16 (39.0%) Complete response: 2 (4.9%) Unknown: 5 (12.2%) Open table in a new tab In our study, no significant differences were found in survival nor in the number of long-term survivors in the group of patients with brain metastases. The activity of immunotherapy on the central nervous system still needs to be defined and the impact on the outcomes of this subgroup of patients requires further investigation.

Nonproportional hazards model with a frailty term for modeling subgroups with evidence of long-term survivors: Application to a lung cancer dataset.

With advancements in medical treatments for cancer, an increase in the life expectancy of patients undergoing new treatments is expected. Consequently, the field of statistics has evolved to present increasingly flexible models to explain such results better. In this paper, we present a lung cancer dataset with some covariates that exhibit nonproportional hazards (NPHs). Besides, the presence of long-term survivors is observed in subgroups. The proposed modeling is based on the generalized time-dependent logistic model with each subgroup's effect time and a random term effect (frailty). In practice, essential covariates are not observed for several reasons. In this context, frailty models are useful in modeling to quantify the amount of unobservable heterogeneity. The frailty distribution adopted was the weighted Lindley distribution, which has several interesting properties, such as the Laplace transform function on closed form, flexibility in the probability density function, among others. The proposed model allows for NPHs and long-term survivors in subgroups. Parameter estimation was performed using the maximum likelihood method, and Monte Carlo simulation studies were conducted to evaluate the estimators' performance. We exemplify this model's use by applying data of patients diagnosed with lung cancer in the state of São Paulo,Brazil.

A Cure Rate Model for Exponentially Distributed Lifetimes with Competing Risks

In real life, more often experimental units are susceptible to more than one risk factor. Moreover, some experimental units may not fail even if they are observed over a long period of time. In statistical analysis, competing risks models handle the first scenario while cure rate models have been introduced to analyze the long-term survivors in the population. In this paper, we consider a cure rate model when the failure of a unit can be due to either of the two competing causes. To analyze the competing risk data in presence of long-term survivors, we consider the latent failure times approach introduced by Cox (J R Stat Soc Ser B (Methodol) 21(2):411–421, 1959). The latent failure times are assumed to follow exponential distributions, and they are independently distributed. Under this setup, a random censoring scheme is applied and the observed data consist of either censored times or actual failure times along with the cause of failures. We derive the maximum likelihood estimators (MLEs) using the expectation-maximization (EM) algorithm based on the missing value principle. As the overall survival function is not a proper survival function, the asymptotic behavior of the MLEs is not immediate. We provide the sufficient conditions for the existence, uniqueness, consistency and the asymptotic normality of the MLEs. Monte Carlo simulations are performed to support the theoretical validation numerically. For illustrative purposes, we have analyzed one real dataset, and the results are quite satisfactory.

Bivariate lifetime models in presence of cure fraction: a comparative study with many different copula functions

In time-to-event studies it is common the presence of a fraction of individuals not expecting to experience the event of interest; these individuals who are immune to the event or cured for the disease during the study are known as long-term survivors. In addition, in many studies it is observed two lifetimes associated to the same individual, and in some cases there exists a dependence structure between them. In these situations, the usual existing lifetime distributions are not appropriate to model data sets with long-term survivors and dependent bivariate lifetimes. In this study, it is proposed a bivariate model based on a Weibull standard distribution with a dependence structure based on fifteen different copula functions. We assumed the Weibull distribution due to its wide use in survival data analysis and its greater flexibility and simplicity, but the presented methods can be adapted to other continuous survival distributions. Three examples, considering real data sets are introduced to illustrate the proposed methodology. A Bayesian approach is assumed to get the inferences for the parameters of the model where the posterior summaries of interest are obtained using Markov Chain Monte Carlo simulation methods and the Openbugs software. For the data analysis considering different real data sets it was assumed fifteen different copula models from which is was possible to find models with satisfactory fit for the bivariate lifetimes in presence of long-term survivors.

Estimating the turning point of the log-logistic hazard function in the presence of long-term survivors with an application for uterine cervical cancer data

The hazard function plays an important role in cancer patient survival studies, as it quantifies the instantaneous risk of death of a patient at any given time. Often in cancer clinical trials, unimodal hazard functions are observed, and it is of interest to detect (estimate) the turning point (mode) of hazard function, as this may be an important measure in patient treatment strategies with cancer. Moreover, when patient cure is a possibility, estimating cure rates at different stages of cancer, in addition to their proportions, may provide a better summary of the effects of stages on survival rates. Therefore, the main objective of this paper is to consider the problem of estimating the mode of hazard function of patients at different stages of cervical cancer in the presence of long-term survivors. To this end, a mixture cure rate model is proposed using the log-logistic distribution. The model is conveniently parameterized through the mode of the hazard function, in which cancer stages can affect both the cured fraction and the mode. In addition, we discuss aspects of model inference through the maximum likelihood estimation method. A Monte Carlo simulation study assesses the coverage probability of asymptotic confidence intervals.

The nonparametric location-scale mixture cure model

We propose completely nonparametric methodology to investigate location-scale modeling of two-component mixture cure models that is similar in spirit to accelerated failure time models, where the responses of interest are only indirectly observable due to the presence of censoring and the presence of long-term survivors that are always censored. We use nonparametric estimators of the location-scale model components that depend on a bandwidth sequence to propose an estimator of the error distribution function that has not been considered before in the literature. When this bandwidth belongs to a certain range of undersmoothing bandwidths, the proposed estimator of the error distribution function is root-n consistent. A simulation study investigates the finite sample properties of our approach, and the methodology is illustrated using data obtained to study the behavior of distant metastasis in lymph-node-negative breast cancer patients.

Semiparametric random censorship models for survival data with long-term survivors

In this article, we study a semiparametric random censorship model for survival data in the presence of long-term survivors. Local likelihood method is employed to estimate the conditional mean regression function of binary random variables. The proposed estimators for the survival function and the cure rate, as well as their asymptotic properties, are investigated based on empirical and U-statistical processes. In particular, the proposed estimator for the cure rate is shown to be superior over the previous estimator considered by Maller and Zhou in the sense of having a smaller asymptotic variance. This semiparametric random censorship model with related estimation methods provide an efficient alternative for survival analysis with long-term survivors.

Increasing incidence of primary central nervous system lymphoma but no improvement in survival in Sweden 2000-2013.

This study aims to characterize the epidemiology of immunocompetent Primary central nervous system lymphoma (PCNSL) diagnosed 2000-2013 in Sweden. Cases were identified in the population-based Swedish Lymphoma Register. Incidence per 100000 person-years and 95% confidence intervals (CI) were calculated, and PCNSL-specific survival was estimated using relative survival. Tests for temporal trends were performed using Poisson regression. Population incidence of all brain tumors was retrieved for comparison. With 359 identified PCNSL cases (median age 66years), overall incidence was 0.26 (95% CI: 0.24-0.29) and the average annual increase 4% (P=.002). The increasing trend was primarily observed among elderly individuals (70+years). Similarly, an increase in incidence of all brain tumors was noted only among the elderly. There was no significant improvement in relative survival across the study period although, among fit patients (with Eastern Cooperative Oncology Group, EGOC 0), survival plateaued 6years after diagnosis. The increasing PCNSL incidence in the elderly was consistent with an increasing incidence of brain tumors of any type and may in part be attributable to improved diagnostics and reporting in this group. New treatment options have not yet translated into general survival improvements in a population-based setting, although the presence of long-term survivors among fit patients is encouraging.

EM algorithm-based likelihood estimation for a generalized Gompertz regression model in presence of survival data with long-term survivors: an application to uterine cervical cancer data

ABSTRACTIn this paper we develop a regression model for survival data in the presence of long-term survivors based on the generalized Gompertz distribution introduced by El-Gohary et al. [The generalized Gompertz distribution. Appl Math Model. 2013;37:13–24] in a defective version. This model includes as special case the Gompertz cure rate model proposed by Gieser et al. [Modelling cure rates using the Gompertz model with covariate information. Stat Med. 1998;17:831–839]. Next, an expectation maximization algorithm is then developed for determining the maximum likelihood estimates (MLEs) of the parameters of the model. In addition, we discuss the construction of confidence intervals for the parameters using the asymptotic distributions of the MLEs and the parametric bootstrap method, and assess their performance through a Monte Carlo simulation study. Finally, the proposed methodology was applied to a database on uterine cervical cancer.

Semiparametric Sequential Testing for Multiple Change Points in Piecewise Constant Hazard Functions with Long-Term Survivors*

In this article, we consider parameter estimation in the hazard rate with multiple change points in the presence of long-term survivors. We combine two methods: maximum likelihood based and martingale based, to estimate the change points in the hazard rate for right censored survival data that accounts for long-term survivors. A simulation study is carried out to compare the performance of estimators. The method is applied to analyze two real datasets.

Statistical issues and challenges in immuno-oncology.

BackgroundThe development of immuno-oncologic agents poses unique challenges, namely that both efficacy and safety profiles differ from previously characterized cytotoxic and pathway-specific agents. In addition, exponential distribution is usually assumed in study designs with time-to-event endpoints such as overall survival or progression-free survival. This assumption might lead to wrong estimates of study duration and statistical power if the phenomena of long term survival and delayed clinical effects are present. The aim here was to evaluate the magnitude of the impact caused by the violation of this assumption, and to describe new ways of analyzing efficacy and safety of immuno-oncologic agents.MethodsMonte Carlo simulation was implemented to explore the impact of long term survivors and delayed treatment effect on study power and trial duration. Scenarios with various combinations of long term and delayed treatment effects were considered. Study power and duration were evaluated based on 10000 randomly generated trial data sets. The utility of group sequential study designs was discussed. A new set of immune-related response criteria (irRC) was considered for efficacy analysis. Two new methods for identifying adverse events, termed immune-related adverse events (irAE) and immune-mediated adverse reactions (imAR) were described. The key features of the safety profiles derived using these two methods were similar. Both methods were aimed at identifying inflammatory adverse events caused by immunotherapies.ResultsThe presence of long term survivors usually lengthened the study duration. Depending on the treatment effect post survival curve separation, delayed clinical effect in general led to a loss of power. The irRC offered a new way of identifying clinical responses. Both safety analyses demonstrated higher sensitivity of identifying adverse events of immune system origin.ConclusionThis simulation study showed the importance of accounting for the delayed treatment effect and long term survivors when these phenomena were expected. Interim analyses for the purpose of stopping the study for either positive or futile outcome should be implemented with caution in immuno-oncology trials. The new efficacy analysis offered a potential new way of assessing signs of activity in immunotherapies. While the irAE method facilitated prompt and effective management of adverse events, the imAR method captured truly immune-related events.

Phase II trial of intracerebral administration of CpG oligonucleotide for patients with recurrent glioblastoma

2043 Background: CpG-ODN are immunostimulating oligodeoxynucleotides containing CpG motifs. When injected locally, they can induce tumor rejection in animal models. In a phase I clinical trial, intra-tumoral infusions of CpG-ODN in glioblastoma patients were well tolerated at doses up to 20 mg. This multicenter phase II trial was designed to study the efficacy and tolerance of a local treatment by CpG-ODN in patients with recurrent glioblastoma. Methods: Patients with recurrent GBM occurring at least three months after radiotherapy and previously treated with one at least regimen of chemotherapy received 20mg CpG- ODN (CpG-28) by convection-enhanced delivery. The percentage of patients without tumour progression at 6 months after inclusion was the primary endpoint. Secondary endpoints were tolerance, survival, and radiological response. Results: Thirty-four patients were enrolled in two centers and thirty-one patients received the treatment. The 6-months progression free survival (PFS) was 19%. One partial response and 3 minor responses were observed. Eight patients (24%) were alive after 1 year, and 5 patients (15%) were alive 2 years after inclusion. The median overall survival was 28 weeks.Treatment was usually well tolerated. Among 6 patients studied, the pharmacokinetic profile of CpG-28 in the blood was heterogeneous. CpG-28 reached up to 79 ng/mL at the end of infusion in one patient. Return to baseline concentration occurred at around 70 hours after the start of infusion. Half-life estimates were variable ranging from 3 to 24 hours. The mean area under the plasma concentration-time curve (AUC) ranged from 0.07 to 0.65 μg x hr/ml. Conclusions: CpG-28 induced a few cases of radiological reponses, but the impact on the 6-months PFS was minimal in this population. However, the presence of long-term survivors suggest that this treatment might have been beneficial in some patients, whose molecular or clinical characteristics remain to be defined. [Table: see text]

Proportional hazards models for survival data with long-term survivors

In this paper we study the Cox proportional hazards model for survival data in the presence of long-term survivors. Both semiparametric and full parametric versions of the Cox model are considered. Partial likelihood and full likelihood are used to obtain the estimators of the coefficients of covariates and the long-term survivor proportion. Their asymptotic properties are also derived based on counting process and martingale theory. Simulations are carried out to check and compare the performance of the estimators between semiparametric and full parametric models.

Biochemotherapy with low doses of subcutaneous interleukin-2 in patients with melanoma: results of a phase II trial

Metastatic melanoma has an ominous prognosis. Bio-chemotherapy regimens can increase the response rate but with a high degree of toxicity due, mainly, to the use of high-dose intravenous interleukin-2. To test the feasibility and activity profile of a bio-chemotherapy regimen of low-dose subcutaneous interleukin-2. Administration scheme: dacarbazine at 200 mg/m2/d on days 1-4, cisplatin at 20 mg/m2/d intravenous on days 1-4, vinblastine at 1.5 mg/m2/d on days 1-4, IL-2 at 4.5 MUI/m2/d subcutaneous on days 5-8, IFN-alpha at 5 MU subcutaneous on days 5-9, 11, 13, 15 of every 21-day cycle. Objective response was obtained in 11 patients (39.3%; 95%CI: 21-59) including 4 with complete response (14.3%; 95%CI: 4-33). With an extended follow-up of 49 months and 60 months, respectively, 2 patients continue with complete response. The main toxicities were haematological: grade 3-4 neutropenia in 8.2% of cycles, thrombocytopenia in 1.2% and anaemia in 3.2%. The regimen is safe and has a good activity profile. The presence of long-term survivors, despite the use of lower doses and subcutaneous IL-2, is encouraging.

Oesophagectomy for squamous cell carcinoma: lessons from a decade of consecutive resections

The aim of this study was to analyse the outcome following oesophageal resection for squamous cell carcinoma (SCC) in a large volume unit. Between 1987 and 1997, 166 patients with SCC underwent oesophagectomy. The outcomes and pathological characteristics of this cohort were then analysed. Operative mortality was 6% (10 patients). Anastomotic leak occurred in 11 (6.6%). A history of previous respiratory disease and anastomotic leak were independent predictors of early mortality (P=0.02). Pathological examination demonstrated the presence of stage I disease in 8, stage IIa in 58, stage IIb in 14 and stage III in 87 patients. Kaplan-Meier survival at 1, 3 and 5 years for all patients was 71.6%, 44.6% and 33.5%. Five-year survival was 87.5% for stage I, 47.1% for stage IIa, 27.4% for stage IIb and 14.5% for stage III. On multivariate analysis, pathological stage (P=0.001) and presence of involved lymph nodes were independent adverse predictors of survival (P<0.0001). In conclusion, oesophagectomy for SCC carries an acceptable risk, which is higher for those having a respiratory disease and those developing an anastomotic leak. The good survival observed in early pathological stages and the presence of long-term survivors amongst those with locally advanced disease are encouraging.