SEARCH · PubMed Health

Results for “Prognostic model”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 163 records · Page 9Linked to original sources

PubMed2002

Multivariate prognostic model for patients with thick cutaneous melanoma: importance of sentinel lymph node status.

BACKGROUND: The overall prognosis of patients with thick cutaneous melanoma (TCM) is generally thought to be poor. Surgically staging these patients with sentinel lymph node (SLN) biopsy remains controversial. This study was performed to determine whether SLN status improved our ability to predict outcome over other known prognostic factors and to develop a model incorporating independent prognostic factors to estimate the risk of recurrence for an individual patient. METHODS: A prospective database identified patients with TCM (>4.0 mm or Clark level V) and clinically negative nodes who underwent SLN biopsy. Univariate and multivariate analyses were performed. RESULTS: From 1991 to 2001, 126 patients were identified; 75 (60%) were male. The median age was 60 years. The median tumor thickness was 5.5 mm, and 43% were ulcerated. Thirty percent of patients had a positive SLN. Recurrence was seen in 50 patients (40%). Median follow-up, relapse-free survival, and overall survival were 25, 50, and 68 months, respectively. Factors independently predictive of recurrence were age >/=60 years, depth >5.5 mm, ulceration, and SLN positivity. SLN status was the most significant prognostic factor (P <.001). The relative risk of recurrence for an individual patient ranged from 1 in patients for whom no adverse factors were present to 29.4 when all factors were present. CONCLUSIONS: SLN status was the strongest independent predictor of outcome in patients with TCM. However, patients with TCM are prognostically heterogeneous, and all independently predictive factors should be considered when an individual patient's risk of recurrence is assessed.

Analysis of Variance↗

PubMed2006

A prognostic model for functional outcome in early rheumatoid arthritis.

OBJECTIVE: To construct a prognostic algorithm to predict 5-year functional outcome in rheumatoid arthritis (RA), based on the Health Assessment Questionnaire (HAQ). METHODS: Data from all patients with 5-year followup (n = 985) were used from an inception cohort, the Early Rheumatoid Arthritis Study (ERAS). Possibly relevant prognostic factors considered in the initial stage of the model-building process were standard clinical, radiological, and laboratory features measured at baseline and at 1 year. Multivariate analysis was performed using logistic regression, and the predictive performance of the model was tested using measures of discrimination and calibration. RESULTS: Bootstrap resampling identified 6 variables that consistently predicted severe functional outcome. Functional grade III/IV (odds ratio 6.7) and HAQ at 1 year (odds ratio 2.4) were the most important. Other variables included socioeconomic status, hemoglobin, and radiographic and disease activity scores. Estimates of the regression coefficients and performance were corrected for over-fitting. Reasonably large values for the c-index (0.82) and the Nagelkerke R(2) (0.39) indicate that the set of prognostic factors explains the variation in outcome to a degree that implies good prediction for individual patients. CONCLUSION: The algorithm identifies patients in the first year of RA who are likely to have poor function by 5 years and who could potentially benefit from aggressive drug therapy. A nomogram is produced for simple application of the model in clinical practice. While further external validation is necessary, this model could allow clinicians to target aggressive therapy earlier in a patient's disease course.

Adult↗

PubMed1992

Development and validation of prognostic models in metastatic breast cancer: a GOCS study.

The significance of several prognostic factors and the magnitude of their influence on response rate and survival were assessed by means of uni- and multivariate analyses in 362 patients with stage IV (UICC) breast carcinoma receiving combination chemotherapy as first systemic treatment over an 8-year period. Univariate analyses identified performance status and prior adjuvant radiotherapy as predictors of objective regression (OR), whereas the performance status, prior chemotherapy and radiotherapy (adjuvants), white blood cells count, SGOT and SGPT levels, and metastatic pattern were significantly correlated to survival. In multivariate analyses favorable characteristics associated to OR were prior adjuvant radiotherapy, no prior chemotherapy and postmenopausal status. Regarding survival, the performance status and visceral involvement were selected by the Cox model. The predictive accuracy of the logistic and the proportional hazards models was retrospectively tested in the training sample, and prospectively in a new population of 126 patients also receiving combined chemotherapy as first treatment for metastatic breast cancer. A certain overfitting to data in the training sample was observed with the regression model for response. However, the discriminative ability of the Cox model for survival was clearly confirmed.

Analysis of Variance↗

PubMed1993

[Katarjian's prognostic model in prognosing chronic myelogenous leukemia at diagnosis and after one year of disease duration].

We estimated survival time of 56 patients with chronic myelogenous leukemia retrospectively classified according to the model by Kantarjian. Moreover we analysed if there was a correlation between the first-year total dose of busulphan and the size of the spleen during the first year of treatment on the one hand and the patients' survival on the other and if the above factors could enhance the prognostic value of the model by Kantarjian. Although we found such a correlation our preliminary results suggest that the first-year total dose of busulphan and the spleen size do not enhance the prognostic value of the model by Kantarjian.

Adolescent↗

PubMed2003

A prognostic model for survival in chronic lymphocytic leukaemia based on p53 expression.

As the abnormal expression of p53 protein is prognostically significant in some human cancers, its significance in patients with B-cell chronic lymphocytic leukaemia (CLL) was assessed. Two investigators evaluated the percentage of bone marrow mononuclear cells that stained for p53, using biopsies stained with anti-p53 monoclonal antibody (DO-7), and graded the degree of staining (0, +, ++, +++). Samples from a cohort of 90 patients with CLL were studied (median age 60 years, range 30-89 years; 57 patients were (63%) previously untreated, 22 patients (24%) had received one or two prior regimens, 11 patients had received (12%) three to seven regimens. The overall percentage of cells positive for p53 staining was a median of 43 (range 1-88). No investigator effect was detected either in overall percentage cells rated p53 positive or on the degree of staining (Pearson's correlation coefficient 0.980, P-value < 0.001). A Cox proportional hazards model showed that the percentage of ++ and +++ p53-positive cells correlated with various prognostic factors in CLL (P < 0.0001). A multivariate model incorporating prior therapy, Rai stage, beta2 microglobulin (beta2M) and p53 expression showed that only the percentage of p53-positive cells and beta2M were predictive of survival, and enabled the development of a highly predictive model of survival based on these two parameters.

Adult↗

PubMed2001

Comparison of prognostic models in patients with advanced Hodgkin disease. Promising results from integration of the best three systems.

BACKGROUND: Several prognostic systems have been elaborated for patients with Hodgkin disease (HD) over the last 12 years, but early identification of a reasonably large group of both low and high risk, advanced stage patients remains unsatisfactory. METHODS: Seven well known models were applied to 516 patients with advanced HD, with 315 patients used for the study sample and 201 patients used for the test sample. Individual performances as well as joint performances were analyzed univariately and multivariately in relation to overall survival, recurrence free survival, and time to treatment failure by means of a proportional hazards model. RESULTS: None of the models identified a group containing > 10% of patients from the total population who had a failure risk of either < or = 10% or > or = 50%. The systems of the International Database on Hodgkin Disease, the Memorial Sloan-Kettering Cancer Center, and the International Prognostic Factor Project showed the best prognostic power; only these three, when analyzed together, predicted clinical outcome with a statistically significant fit to the clinical data. Integration of the three systems in a linear model dramatically improved their individual discriminatory capacity by identifying patients with 10% and 50% failure risks, respectively, in 23% and 24% of the study patient population and in 19% and 25% of the test population, respectively. CONCLUSIONS: As powerful and simple new prognostic factors are awaited that may improve our predictive ability, this integrated index is probably the best way to exploit the significance of those presently available. The program required for the calculations can be downloaded from the Internet at the web site http://www.unimo.it/gisl/default.htm.

Adolescent↗

PubMedDate not supplied

A primer of biostatistic and economic methods for diagnostic and prognostic modeling in nuclear cardiology: Part I.

This review presents a brief overview of existing diagnostic and prognostic methodologies to be used for the evaluation of patients undergoing noninvasive testing. In part I of this review, we will present methods for use of logistic and Cox regression analyses in determining the diagnostic and prognostic value of nuclear imaging techniques. In part II of this review, we will present an outline for the integration of economic evaluations into the clinical decision-making process. This review will document how cost estimates may be defined as well as the differing type of cost analyses (i.e., cost efficiency versus cost-effectiveness).

Coronary Disease↗

PubMed2026

A stratified urine-based molecular diagnostic and prognostic model for non-muscle-invasive bladder cancer management.

BACKGROUND: Non-muscle-invasive bladder cancer (NMIBC) is characterized by a high recurrence rate requiring lifelong cystoscopic surveillance. Existing urine-based molecular assays mainly rely on mutations or methylation, which fail to capture large-scale genomic instability. Copy number variation (CNV) profiling offers complementary information on tumor evolution and aggressiveness, but its application in urinary diagnosis remains limited. We aimed to integrate CNV and DNA methylation signals from urinary DNA to establish a noninvasive and biologically informed stratified diagnostic model for NMIBC recurrence surveillance and risk stratification. METHODS: Urine samples were prospectively collected from 91 patients (75 evaluable) between June 2021 and August 2023. Shallow whole-genome sequencing (sWGS) was used to detect CNVs at chromosomal arm and focal gene levels, while ONECUT2 promoter methylation was quantified by qPCR. Diagnostic and prognostic performance was evaluated by ROC analysis, Kaplan-Meier survival, and stratified recurrence assessment. RESULTS: We evaluated a stratified diagnostic model combining CNV and ONECUT2 methylation testing in a cohort of 79 patients. CNV analysis alone showed high specificity (0.923) for NMIBC diagnosis. A combined model, using CNV as an initial screen followed by ONECUT2 methylation testing in CNV-positive cases, achieved a sensitivity of 0.783, specificity of 0.981, and a negative predictive value (NPV) of 0.911. This approach reduced the number of required ONECUT2 tests by 35% and identified a high proportion of true-negative patients (98.1%), which may help reduce unnecessary cystoscopy procedures. The model also demonstrated significant prognostic value, with the molecularly defined high-risk group showing significantly shorter recurrence-free survival (RFS) than the low-risk group (median RFS: 4.33 months vs. not reached; p < 0.001). Additional, in patients with initially negative cystoscopy after urine sample collection, the model demonstrated a predictive accuracy of 0.922 for recurrence, with molecular positivity observed a median of 9.6 months prior to clinical diagnosis. CONCLUSIONS: Integrating CNV and DNA methylation profiling from urinary DNA provides a powerful and noninvasive molecular framework for NMIBC surveillance. By combining early epigenetic changes with genomic instability signals, this approach enhances recurrence risk assessment and enables earlier detection compared with conventional cystoscopy. It offers a practical route toward personalized and adaptive post-treatment monitoring of NMIBC. TRIAL REGISTRATION: NCT04994197.

Humans↗

PubMed2000

Identifying and modelling prognostic factors with censored data.

A major issue in the analysis of diseases is the identification and assessment of prognostic factors relevant to the development of the illness. Statistical analyses within the proportional hazards framework suffer from a lack of flexibility due to stringent model assumptions such as additivity and time-constancy of effects. In this paper we use tree-based models and varying coefficient models to allow for detectability of prognostic factors with possibly non-additive, non-linear and time-varying impact on disease development. Questions concerning model and smoothing parameter selection are addressed. An analysis of a data set of breast cancer patients demonstrates the ability of these methods to reveal additional insight into the disease influencing mechanisms.

Adult↗

PubMed2000

Short-term prognosis in severe adult and adolescent malnutrition during famine: use of a simple prognostic model based on counting clinical signs.

CONTEXT: In the setting of famine, infection is likely to cause mortality among severely malnourished persons. Although clinical signs are likely to be useful prognostic indicators in this setting, use of a clinical assessment model has not been studied. OBJECTIVE: To examine the use of clinical signs in the prediction of short-term mortality in severely malnourished adults and adolescents during famine. DESIGN: Retrospective cohort study. SETTING: Concern Worldwide Adult Therapeutic Feeding Center in Baidoa, Somalia. PATIENTS: Data from the clinical records of 383 adult and adolescent inpatients admitted to the center between November 1992 through March 1993 who were aged 15 years or older and had a body mass index (BMI) of 13.5 kg/m(2) or less or any signs of edematous malnutrition. MAIN OUTCOME MEASURES: Association of mortality with presence or absence of 8 clinical signs (edema, hydration, ascites, dysentery, diarrhea, anemia, chest infection, and ability to stand) and BMI at admission, and sensitivity and specificity of models including a count of clinical signs and BMI in the prediction of mortality at the center. RESULTS: Ninety-one patients (23.8%) died, with a median time to death of 8 days from admission. Of the 8 clinical signs, severe edema (unadjusted odds ratio [OR], 2.45; 95% confidence interval [CI], 1.41-4.27), apparent dehydration (unadjusted OR, 2. 73; 95% CI, 1.60-4.66), and inability to stand (unadjusted OR, 2.96; 95% CI, 1.40-6.26) were independently associated with mortality. The most useful clinical model was that based on the presence of any 1 of these 3 signs, with a sensitivity of 77% and a specificity of 59%. Ability of admission BMI to predict mortality was less than random. CONCLUSIONS: Models based on clinical signs predicted death better than BMI. Simple counts of clinical signs performed as well as more complex models based on the addition of ORs. Counting relevant clinical signs is an easy and effective prognostic tool in severe adult and adolescent malnutrition during famine; however, it is not sensitive enough for use as a screening tool. JAMA. 2000;284:621-626

Adolescent↗

PubMed1994

Monoclonal antibody BrE-3 participation in a multivariate prognostic model for infiltrating ductal carcinoma of the breast.

Monoclonal antibody (MoAb) BrE-3, an anti-human milk fat globule (HMFG) MoAb, is used here as a novel prognostic indicator for survival and relapse time in patients with infiltrating ductal carcinoma of the breast. A scoring system (4-Score method) was developed to this effect that measured, in a statistically reliable fashion, the level of expression of the epitope for MoAb BrE-3 in the cytoplasm and membranes of breast carcinoma cells in paraffin-embedded sections. In univariate analysis, data obtained by the 4-Score Method as well as data from traditional prognostic indicators (tumor size, axillary node status, and grade of differentiation) were found to be associated with patient survival and relapse. In multivariate analysis, using a Cox proportional hazards regression model, levels of expression of BrE-3 epitope plus tumor size and axillary node status were weighted and combined in an Individual Linear Composite Prognostic Score (ILCPS) that had a high level of association with survival and relapse time in this sample model of patients with infiltrating ductal carcinoma of the breast. This level of association was found to be higher than the level of association for any other combination of traditional or 4-Score method variables. The level of expression of BrE-3 significantly adds to the prognostic capacity of traditional prognostic markers for infiltrating ductal carcinoma of the breast.

Adult↗

PubMed2006

A new prognostic model for FIGO stage 1 epithelial ovarian cancer.

BACKGROUND: No consensus exists which patients with surgical stage 1 epithelial ovarian should receive postoperative chemotherapy. The purpose of this study was to evaluate the prognostic impact of preoperative CA-125 and to establish a prognostic index to identify patients in different risk categories. METHODS: Data of 600 surgically staged patients with FIGO stage 1 EOC treated in eleven gynecological cancer centers in Australia, the USA and Europe were analyzed. Eligible patients include those with invasive EOC where a preoperative CA-125 was obtained and standard surgical staging performed. Overall survival (OS) was chosen as study endpoint. Preoperative CA-125 values were compared with other prognostic factors, and univariate and multivariate Cox models were calculated. RESULTS: Two hundred and one patients (33.5%) had preoperative CA-125 < or =30 U/ml and CA-125 levels < or =30 U/ml were associated with lower grade, sub-stage 1A and mucinous histologic cell type. Patients with elevated CA-125 levels were more likely to receive chemotherapy. OS probability was 95% and 85% for patients with pretreatment CA-125 < or =30 U/ml and >30 U/ml, respectively (p 0.003). Multivariate analysis confirmed preoperative serum CA-125 >30 U/ml (OR 2.7) and age at diagnosis >70 years (OR 2.6) as the only independent predictors for overall survival. CONCLUSION: Pretreatment of CA-125 < or =30 U/ml dominates over histologic cell type, sub-stage and grade to identify a subgroup of FIGO stage 1 patients with a genuinely good prognosis with extremely good survival and who could possibly be spared with adjuvant chemotherapy.

Adult↗

PubMed1993

Results of fludarabine and prednisone therapy in 264 patients with chronic lymphocytic leukemia with multivariate analysis-derived prognostic model for response to treatment.

Two hundred sixty-four patients with chronic lymphocytic leukemia were treated with fludarabine 30 mg/m2 intravenously for 30 minutes each day for 5 days and with prednisone 30 mg/m2 orally each day for 5 days. Courses were repeated monthly. Of the 264 patients. 125 patients (47%) had Rai stage III-IV disease; 169 patients (64%) were previously treated with a median of 3 prior regimens; and 138 of them (82%) were refractory to therapy with alkylating agents. The overall response (OR) and complete response (CR) rates in the 169 previously-treated patients were 52% and 37%; these were 74% and 63%, respectively, in Rai stage O-II patients and declined to 64% and 46%, respectively, in Rai III-IV disease. Among the previously untreated patients, the OR and CR rates were 79% and 63%, these being 85% and 70%, respectively, in Rai O-II patients, and declining to 64% and 46%, respectively, in Rai III-IV disease. The incidence of minor infections or fever of unknown origin was similar in all patient groups and occurred in 22% of courses. The incidence of sepsis and/or pneumonia was significantly correlated with the extent of prior therapy and with Rai stage, and ranged from 3% of courses in the previously untreated Rai O-II patients, to 13% of courses in the previously treated Rai III-IV patients. Listeria sepsis or Pneumocystis carinii pneumonia was noted in 14 patients. With therapy, CD4 levels were uniformly depressed from a median 1,015/microL pretreatment to a median 159/microL after 3 months of fludarabine therapy. Median time to progression in previously treated patients was 22 months. In previously untreated patients, median time to progression was 30 months for patients who achieved a partial remission and has not been reached in patients who achieved a CR with a median follow-up of 2 years. The median survival was 18 months for previously treated patients and has not been reached for previously untreated patients. Response rates in previously treated and untreated patients, as well as infection rates, were identical to those seen in 110 patients treated with the same dose schedule of fludarabine alone. Logistic regression analysis selected 4 factors to be significantly associated with worse response: Rai III-IV stage disease, prior therapy, older age, and low albumin levels. The regression equation was used to derive a probability of response based on the 4 characteristics. When the model was applied to the same population, patients could be divided into 4 prognostic groups with different outcomes.

Antigens, CD↗

PubMed2000

Treatment of small hepatocellular carcinoma with percutaneous ethanol injection: a validated prognostic model.

OBJECTIVE: Percutaneous ethanol injection may prolong the survival of patients with small hepatocellular carcinoma associated with cirrhosis. The aim was to identify prognostic factors of survival and of local recurrence, as well as separate new lesions. METHODS: We performed Cox regression analysis in 115 consecutive patients with hepatocellular carcinoma (81 Child-Pugh class A, 34 Child-Pugh class B) treated by percutaneous ethanol injection. The validity of the model was tested by comparing predicted and observed survival in 105 independent patients from an external series. RESULTS: Overall survival rates were 89%, 63%, and 43% at 1, 2, and 3 yr, respectively. The 1-, 2-, and 3-yr survival rates were 96%, 78%, and 63%, respectively, for Child-Pugh class A patients and were 73%, 35%, 12%, respectively, for Child-Pugh class B. The albumin level was the only independent variable significantly associated with survival (p < 0.0001). The 3-yr rate of appearance of separate new lesions and local recurrence were 41% and 23%, respectively. The survival predicted by the model agreed with that observed in the independent patients. CONCLUSIONS: Survival of patients with hepatocellular carcinoma treated by percutaneous ethanol injection is related to baseline albumin level. The high rate of recurrence (both local and distant) points out the palliative role of this therapy.

Aged↗

PubMed2003

How to assess prognostic models for survival data: a case study in oncology.

OBJECTIVES: A lack of generally applicable tools for the assessment of predictions for survival data has to be recognized. Prediction error curves based on the Brier score that have been suggested as a sensible approach are illustrated by means of a case study. METHODS: The concept of predictions made in terms of conditional survival probabilities given the patient's covariates is introduced. Such predictions are derived from various statistical models for survival data including artificial neural networks. The idea of how the prediction error of a prognostic classification scheme can be followed over time is illustrated with the data of two studies on the prognosis of node positive breast cancer patients, one of them serving as an independent test data set. RESULTS AND CONCLUSIONS: The Brier score as a function of time is shown to be a valuable tool for assessing the predictive performance of prognostic classification schemes for survival data incorporating censored observations. Comparison with the prediction based on the pooled Kaplan Meier estimator yields a benchmark value for any classification scheme incorporating patient's covariate measurements. The problem of an overoptimistic assessment of prediction error caused by data-driven modelling as it is, for example, done with artificial neural nets can be circumvented by an assessment in an independent test data set.

Breast Neoplasms↗

PubMed2002

Prognostic model for relapse after high-dose chemotherapy with autologous stem-cell transplantation for stage IV oligometastatic breast cancer.

PURPOSE: To study prognostic factors after high-dose chemotherapy (HDC) for patients with stage IV oligometastatic breast cancer. PATIENTS AND METHODS: Sixty patients with minimal metastatic disease amenable to local therapy enrolled onto a prospective HDC trial were analyzed for potential prognostic factors. Tumor blocks were retrospectively collected from referring institutions. RESULTS: Median follow-up was 62 months (range, 4 to 120 months). Median relapse-free survival (RFS) and overall survival (OS) times were 52 and 80 months, respectively. Five-year RFS and OS rates were 52% (95% confidence interval [CI], 39% to 64%) and 62% (95% CI, 49% to 74%), respectively. HER-2 expression, number of tumor sites, primary axillary nodal ratio (number of positive nodes divided by number of sampled nodes), number of positive axillary nodes, and delivery or omission of radiotherapy to metastases correlated with RFS. HER-2 overexpression and more than one site were independent adverse risk factors for RFS. HER-2 and the axillary nodal ratio were independent predictors of OS. The following prognostic categories for RFS were established (RFS rate, median RFS): good risk, no factors (77%, 80 months); intermediate risk, one factor (41%, 28 months); and poor risk, both factors (10%, 10 months). CONCLUSION: Long-term results in patients with oligometastatic breast cancer are encouraging but need validation in prospective randomized studies. HER-2 expression, number of sites, and primary nodal ratio are independent outcome predictors. Confirmation of these observations in this selected population would imply the need for reevaluation of the current tenet that early detection of metastatic breast cancer recurrence is of no benefit.

Adult↗

PubMed2006

Splenic marginal zone lymphoma: a prognostic model for clinical use.

The Integruppo Italiano Linfomi (IIL) carried out a study to assess the outcomes of splenic marginal zone lymphoma and to identify prognostic factors in 309 patients. The 5-year cause-specific survival (CSS) rate was 76%. In univariate analysis, the parameters predictive of shorter CSS were hemoglobin levels below 12 g/dL (P < .001), albumin levels below 3.5 g/dL (P = .001), International Prognostic Index (IPI) scores of 2 to 3 (P < .001), lactate dehydrogenase (LDH) levels above normal (P < .001), age older than 60 years (P = .01), platelet counts below 100,000/microL (P = .04), HbsAg-positivity (P = .01), and no splenectomy at diagnosis (P = .006). Values that maintained a negative influence on CSS in multivariate analysis were hemoglobin level less than 12 g/dL, LDH level greater than normal, and albumin level less than 3.5 g/dL. Using these 3 variables, we grouped patients into 3 prognostic categories: low-risk group (41%) with no adverse factors, intermediate-risk group (34%) with one adverse factor, and high-risk group (25%) with 2 or 3 adverse factors. The 5-year CSS rate was 88% for the low-risk group, 73% for the intermediate-risk group, and 50% for the high-risk group. The cause-specific mortality rate (x 1000 person-years) was 20 for the low-risk group, 47 for the intermediate-risk group, and 174 for the high-risk group. This latter group accounted for 54% of all lymphoma-related deaths. In conclusion, with the use of readily available factors, this prognostic index may be an effective tool for evaluating the need for treatment and the intensity of therapy in an individual patient.

Adult↗

PubMed1986

Tumor burden assessment and its implication for a prognostic model in advanced diffuse large-cell lymphoma.

Previously untreated adult patients who presented with advanced diffuse large-cell lymphoma (DLCL) at diagnosis were studied to identify possible prognostic factors. One hundred five patients were seen between 1974 and 1981; 45 patients were stage III and 60 patients were stage IV. All patients received cyclophosphamide, doxorubicin, vincristine, prednisone, and bleomycin (CHOP-Bleo). Stage III patients also received radiation therapy alternated with chemotherapy. Overall survival was 50% at 5 years and 43% at 8 years. Seventy-four patients achieved a complete remission (CR) and 37 are alive and disease-free with a median follow-up of 72 months. There was no difference in clinical outcome between stage III and stage IV. However, a proportional hazards model identified lactic dehydrogenase (LDH) level and tumor burden, among all clinical factors studied, as independent risk factors for survival. These two factors were also important for achievement of remission and relapse-free survival. Three distinct patient risk groups were identified with 5-year survival rates of 87%, 48%, and 20%, respectively. The measure of tumor burden proposed herein, along with LDH level, can be used for developing treatment programs, and for meaningful comparison of different treatment regimens, as well as assessment of prognosis.

Abdominal Neoplasms↗