Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Multivariate Analysis”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

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

At least 127 records · Page 7Linked to original sources
PubMed2000

Prognostic indicators for renal cell carcinoma: a multivariate analysis of 643 patients using the revised 1997 TNM staging criteria.

PURPOSE: We determine independent prognostic indicators for renal cell carcinoma using the revised 1997 TNM staging criteria. MATERIALS AND METHODS: The records of 643 consecutive patients undergoing partial or radical nephrectomy at our institution between 1987 and 1998 were reviewed. Preoperative evaluation of functional status using the Eastern Cooperative Oncology Group (ECOG) criteria was performed in all cases. Renal cell carcinoma grade and stage were evaluated using the 1997 American Joint Committee on Cancer grading and TNM staging criteria, respectively. Patients were followed for a mean plus or minus standard deviation of 47+/-40 months (median 87). Kaplan-Meier survival curves were used to determine 5-year cancer specific survival for all patient groups. Univariate analysis using log rank sum tests was performed to evaluate the prognostic significance of overall TNM stage, tumor stage, disease grade and ECOG status. Multivariate analysis was performed to determine which factors had an independent impact on survival of patients with renal cell carcinoma. RESULTS: The 5-year cancer specific survival rate was 91%, 74%, 67% and 32% for TNM stages I, II, III and IV lesions, respectively (p<0.001). Analysis demonstrated a survival rate of 83% for stage T1, 57% for stage T2, 42% for stage T3 and 28% for stage T4 disease (p<0.001), and 89% for grade 1, 65% for grade 2, and 46% for grades 3 and 4 (p<0.001). Multivariate analysis revealed that overall TNM stage and grade of disease were the most important prognostic indicators for renal cell carcinoma (p<0.001). ECOG classification was a less significant predictor (p = 0.031) and tumor stage was not shown to have any independent impact on patient survival (p = 0.138). CONCLUSIONS: Better survival rates of patients with localized and advanced renal cell carcinoma can be demonstrated with recent advances in diagnosis and treatment. The revised 1997 TNM criteria manifest an appropriate adjustment in staging renal cell carcinoma based on these improvements, with overall stage correlating with cancer specific survival. In contrast, while effectively predicting survival, tumor stage did not demonstrate an independent impact on renal cell carcinoma prognosis under multivariate analysis. Instead, other factors, such as ECOG status and more importantly grade of disease, appeared to affect survival significantly as independent elements. Based on our recent experience with patients treated for renal cell carcinoma in the era of enhanced technology and improved survival, tumor grade and molecular markers may serve as useful adjuncts to TNM staging in guiding treatment and predicting survival outcomes.

Adult↗
PubMed1994

Prognostic value of MDR1 gene expression in neuroblastoma: results of a multivariate analysis.

The prognostic value of the MDR1 gene expression in neuroblastoma (NB) was assessed in a multivariate analysis performed in a series of 84 patients (pts) taking into account the main known clinical and biological factors of the disease, i.e., age, stage, MYCN genomic content and DNA ploidy index. Twenty seven children were < 1 year (yr), 13 presented with stage I and II, 7 with stage IV-S, 17 with stage III and 47 (56%) with stage IV. Tumor specimens were obtained from involved bone marrow (n = 12) or surgical primary tumor specimens (n = 72). MDR1 gene expression was measured by Northern hybridization technique and expressed in arbitrary units (a. u.) (Goldstein et al., 1989). Analysis of MYCN genomic content and DNA ploidy index were performed by Southern blot hybridization technique and flow cytometry, respectively. Out of 84 tumor specimens 19 (23%) showed MYCN amplification (> 3 copies/haploid genome). In 24 cases (29%) no detectable MDR1 gene transcript was found (0 a.u.) whereas 42 (50%) had a value in the range 1-30 a.u., and 18 (21%) a value beyond 30 a.u.. High transcript levels were found in localized as well as in metastatic NB (NS). No significant correlation between MDR1 gene expression, age, stage, or MYCN genomic content was found In univariate analysis stage IV, age > 1 yr, MYCN amplification, diploid DNA content and high MDR1 gene transcript levels were significantly related to an increased risk of death. In multivariate analysis only stage IV, MYCN amplification and MDR1 overexpression remained significantly associated with an increased risk of death.(ABSTRACT TRUNCATED AT 250 WORDS)

ATP Binding Cassette Transporter, Subfamily B, Mem↗
PubMed1998

Prognostic factors in Sézary syndrome: a multivariate analysis of clinical, haematological and immunological features.

BACKGROUND: Sézary syndrome (SS) prognostic factors are not well defined because of the rarity of this disease. The specific goal of this prospective study was to assess by multivariate analysis the predictive value with respect to survival of a series of clinical, haematological and immunological parameters taken at SS diagnosis. PATIENTS AND METHODS: A cohort of 62 SS patients diagnosed and followed since 1975 was examined, and 51 were included in the multivariate analysis model. RESULTS: The median survival time was 31 months (range: 1 month-15.7+ years), and the five-year survival rate 33.5%. The following variables were found by univariate analysis to be associated with a poor prognosis at the time of SS diagnosis: previous history of mycosis fungoides (P = 0.013), high number of circulating leukocytes (P = 0.001), Sézary cells (SC) (P < 0.001) and CD4+ cells (P < 0.001), presence of large circulating SC (P < 0.001), above normal range LDH serum levels (P = 0.015), presence of PAS-positive inclusions in the cytoplasm of circulating SC (P < 0.001), high CD4/CD8 ratio (P = 0.004) and a CD7 negative circulating SC phenotype (P < 0.001). Among them, the stepwise multivariate analysis selected as adverse independent prognostic factors: PAS-positive cytoplasmic inclusions (P = 0.001), CD7 negative phenotype (P = 0.018) and presence of large circulating SC (P = 0.045). CONCLUSIONS: Two low-/high-risk groups have been singled out on the basis of the risk index. Patients with no or one adverse prognostic feature(s) (risk index < or = 1; n = 31) share a slow disease course and a relatively favorable prognosis (five-year survival: 58%); on the other hand, patients with 2 or 3 adverse prognostic feature (risk index > 1; n = 20) are characterized by an aggressive disease course not modifiable by traditional therapies (five-year survival: 5%).

Adult↗
PubMed2004

Predictors of prostate cancer on extended biopsy in patients with high-grade prostatic intraepithelial neoplasia: a multivariate analysis model.

OBJECTIVE: To define the importance of extended biopsy in patients with high-grade prostatic intraepithelial neoplasia (HGPIN) and to define predictors of cancer in extended biopsy in patients with HGPIN, using multivariate analysis. PATIENTS AND METHODS: In all, 83 patients with previous sextant biopsy of HGPIN had an extended 11-core biopsy taken. Patients with a negative biopsy for cancer were followed by serum prostate-specific antigen (PSA) and digital rectal examination (DRE) every 6 months. The extended biopsy was repeated in 21 patients. The criteria for second biopsy were an increase in PSA and/or abnormal changes on DRE. Overall, 49 patients had a transurethral resection of the prostate (TURP). The cancer-detection rate on extended biopsy was correlated with risk factors using the chi-square test and multivariate analysis. RESULTS: Extended biopsy detected prostate cancer in 30 of the 83 men (36%), with positive cores in only 20 sextant biopsy sites (67%), in only seven in additional sites (23%), and both in three (10%). Of the 21 patients who had repeat extended biopsy, four (19%) had cancers. There were two carcinomas in the 49 TURP specimens (4%). The PSA level, DRE and transrectal ultrasonography findings were not predictive of cancer in extended biopsies (chi-square test). Patient age, PSA density and the number of cores with HGPIN (all P < 0.001) had a significant effect on the cancer-detection rate, and multivariate analysis showed that all three were independent predictors of cancer. A logistic regression model was designed to predict the probability of cancer in extended biopsy, with an overall accuracy of 78%. CONCLUSION: Extended biopsy improved the cancer detection rate by 23% in patients with HGPIN. Patient age, PSA density and the number of cores with HGPIN were the only independent predictors of cancer.

Aged↗
PubMed2003

Classification of lactate dehydrogenase of different origin by liquid chromatography-mass spectrometry and multivariate analysis.

A method intended to serve as a multivariate quality control tool in the production of pharmaceutical proteins is presented. The method is based on multivariate analysis of peptide maps generated with liquid chromatography-mass spectrometry (LC-MS). Lactate dehydrogenase (LDH) from different species and tissues were used as model compounds in the study. The proteins were digested with Endoproteinase Lys-C before the LC-MS analysis. After data pretreatment of the peptide maps, successful classification of the LDHs were obtained by discriminant analysis with partial least squares regression and artificial neural networks. Further, principal component analysis was applied to visualize the relationships between the samples.

Animals↗
PubMed1992

[Multivariate analysis of the clinical signs in late cortical cerebellar atrophy (LCCA) in Japan--compared with olivo-ponto-cerebellar atrophy (OPCA) and hereditary cortical cerebellar atrophy (HCCA)].

We investigated the clinical features of 179 patients with late cortical cerebellar atrophy (LCCA) comparing with 382 patients with olivo-ponto-cerebellar atrophy (OPCA) and 91 patients with hereditary cortical cerebellar atrophy (HCCA) using multivariate analysis. They had no significant difference in durations from onset. Disorders of finger-nose test were of more severity and disorders of heel-knee test were less involved in LCCA than in HCCA. Eye movements tended to be normal in LCCA. LCCA showed more depressive and euphoric character than OPCA and HCCA. Analysis of LCCA and OPCA demonstrated that LCCA revealed hypotonia in muscles, normal or weak reactions of muscle stretch reflexes. There existed 21.4 percent of cases which could not be discriminated by multivariate analysis using clinical signs in LCCA and OPCA. Multivariate analysis of LCCA and HCCA, which are very similar in clinical signs, showed 66.3 percent of discriminatory rate. Thus, we could diagnose them to a certain extent only by clinical signs.

Aged↗
PubMed1994

Lipiodolization for unresectable hepatocellular carcinoma: an analysis of 205 patients using univariate and multivariate analysis.

Since 1982 we have developed and successfully performed lipiodolization on 205 patients with unresectable hepatocellular carcinoma. The 1-, 2-, 3-, 4-, and 5-year survival rates were 55.6%, 31.7%, 16.3%, 8.7%, and 2.9%, respectively, while the median survival was 413 days, and 20 patients survived longer than 3 years after the first lipiodolization. Our study demonstrated that the factors for a comparatively good prognosis were as follows: patient older than 50 years, nodular type hepatocellular carcinoma < 5 cm in diameter, negative tumor invasion of portal vein, < 10 intrahepatic metastases, serum alpha-fetoprotein level < 400 ng/ml, and Okuda stage I. Lipiodolization did not cure hepatocellular carcinoma. It did, however, in this study, achieve a prolongation of the patient's survival.

Adult↗
PubMed1982

Numerical evaluation of cytologic data. XI. Nested designs in multivariate analysis of variance.

Nested designs in multivariate analysis of variance allow the investigator to assess at what stages in an experimental procedure variability enters and to what extent. When employed in conjunction with a fixed-effects level in the design, a mixed model results, which provides the appropriate significance tests for the fixed-level effect; e.g., differences in cells between control patients and treated patients must be tested against the patient-to-patient variability, not against the cell-to-cell variability. A worked example for a three-level, mixed-model nested design is given, including the significance tests.

Analysis of Variance↗
PubMed1989

[Outbreaks of asthma attacks and meteorological parameters--multivariate analysis].

The correlation between patient visits and meteorological parameters was analyzed among asthmatic children who visited the out-patient emergency clinic at the Teikyo University Hospital during the three years from 1984 to 1986. Data were analyzed with the method of multivariate analysis (quantification theory type II). The major factors affecting the frequency of visits were an air temperature at more than 15 degrees C, a relative humidity at more than 65%, a vapor pressure at more 1013 mb and a wind speed at less than 3.5 m/sec. Multivariate analysis of seven meteorological parameters, the four parameters shown above plus sea level pressure, cloud density and precipitation, revealed that discrimination ratios with the Lag 3, Lag 2, Lag 1 and Lag 0 values were 63.8%, 68.5%, 68.5% and 63.6%, respectively. These results indicate that asthma forecast is possible by the multivariate analysis of meteorological conditions.

Asthma↗
PubMed1999

Determination of c-myc amplification and overexpression in breast cancer patients: evaluation of its prognostic value against c-erbB-2, cathepsin-D and clinicopathological characteristics using univariate and multivariate analysis.

C-myc and c-erbB-2 amplification and/or overexpression as well as total cathepsin-D (CD) concentration have been reported to be associated with poor prognosis in breast cancer. The prognostic significance, however, remains somewhat controversial, partly because of discrepancies among the different methodologies used. We determined the amplification and overexpression of c-myc oncogene in 152 breast cancer patients and examined its prognostic value in relation to c-erbB-2 amplification and overexpression, high concentration of CD (> or = 60 pmol mg(-1) protein) and standard clinicopathological prognostic factors of the disease. High CD concentration, as well as c-myc amplification and overexpression, proved to be the best of the new variables examined for prediction of early relapse (ER; before 3 years). After multivariate analysis only CD remained significant, which suggests that the prognostic power of these variables is similar. Using univariate analysis we proved that c-myc amplification and overexpression were highly significant for disease-free survival (DFS) (P = 0.0016 and P = 0.0001 respectively) and overall survival (OS) (P < 0.0001 and P = 0.0095 respectively), although by multivariate analysis c-myc overexpression was statistically significant only for DFS (P = 0.0001) and c-myc amplification only for OS (P = 0.0006). With regard to c-erbB-2, only its overexpression appeared to be significant for DFS and OS, although after multivariate analysis its prognostic power was weaker (P = 0.030 and P = 0.024 respectively). c-myc amplification and overexpression exhibited a tendency for locoregional recurrence (LRR) (P = 0.0024 and P = 0.0075 respectively), however, their prognostic value was lower after multivariate analysis and only CD remained significant.

Adult↗
PubMed2004

Metabolomic analysis of Strychnos nux-vomica, Strychnos icaja and Strychnos ignatii extracts by 1H nuclear magnetic resonance spectrometry and multivariate analysis techniques.

1H Nuclear magnetic resonance spectrometry and multivariate analysis techniques were applied for the metabolic profiling of three Strychnos species: Strychnos nux-vomica (seeds, stem bark, root bark), Strychnos ignatii (seeds), and Strychnos icaja (leaves, stem bark, root bark, collar bark). The principal component analysis (PCA) of the 1H NMR spectra showed a clear discrimination between all samples, using the three first components. The key compounds responsible for the discrimination were brucine, loganin, fatty acids, and Strychnos icaja alkaloids such as icajine and sungucine. The method was then applied to the classification of several "false angostura" samples. These samples were, as expected, identified as S. nux-vomica by PCA, but could not be clearly discriminated as root bark or stem bark samples after further statistical analysis.

Magnetic Resonance Spectroscopy↗
PubMedDate not supplied

Multivariate analysis of determinants of postoperative wound infection: a possible basis for intervention.

Multivariate analysis of variables hypothetically associated with the development of postoperative wound infection was attempted with use of large groups of patients undergoing certain surgical procedures over periods of 12-36 months. The data were obtained through prospective surveillance by specially trained nurse-epidemiologists. For patients who had colon surgery (n=261), the multivariate analysis showed that only four variables were independently associated with postoperative wound infection: more than one operation during a single episode of hospitalization (adjusted odds ratio, 7.3); Arab ethnicity (adjusted odds ratio, 6.1); prophylaxis according to the recommended protocol not given (adjusted odds ratio, 2.4); and opening of a colostomy as the specific operation (adjusted odds ratio, 2.3). Of all the variables analyzed, more than one operation during an admission was consistently associated with the highest risk of infection in patients who had cardiac and neurosurgery as well as colon surgery. On the other hand, a traditional risk factor, e.g., the presence of drains at the site of the operation, was not found to be independently associated with an increased risk of infection in any of the groups tested. This type of analysis is essential to facilitate the introduction of a more meaningful program for intervention based on proved rather than hypothetical risk factors

Analysis of Variance↗
PubMed1995

The impact of the inclusion of endorectal coil magnetic resonance imaging in a multivariate analysis to predict clinically unsuspected extraprostatic cancer.

BACKGROUND: The introduction of the endorectal coil magnetic resonance imaging (MRI) technique has improved the accuracy of preoperative staging for prostate cancer. This study quantifies the improvement in the ability to identify clinically unsuspected extraprostatic disease with the use of the endorectal coil MRI. METHODS: A retrospective review of the pathologic findings of 347 patients with prostate cancer treated with a radical retropubic prostatectomy was performed. The preoperative clinical indicators including prostate specific antigen (PSA), clinical stage, Gleason score, and endorectal coil MRI data were employed in a multivariate analysis to identify patients who were at high risk for seminal vesicle invasion (SVI) or extracapsular extension (ECE). The sensitivity, specificity, and positive and negative predictive values for predicting SVI and ECE were calculated using the significant clinical indicators found on the multivariate analysis. RESULTS: The clinical factors identified on multivariate analysis as significant predictors of SVI include the endorectal coil MRI data (P < 0.0001), PSA (P = 0.0096), and the Gleason score (P = 0.012). Endorectal coil MRI data (P < 0.0001), PSA (P = .0001), and Gleason score (P < .0001) were significant predictors of ECE. In the patient subgroup with PSA (> 10-20 ng/ml) and Gleason score of 5 to 7, the addition of the endorectal coil MRI data enabled an additional 71 and 27% of patients with SVI and ECE, respectively, to be correctly identified. These patients would have been missed based on the prediction obtained from the PSA and Gleason score alone. CONCLUSIONS: The use of the endorectal coil magnetic resonance imaging data, in addition to prostate specific antigen and Gleason score, provides a more accurate prediction of the pathologic outcome of seminal vesicle invasion and extracapsular extension than the PSA and Gleason score alone for the patient subgroup with a PSA of greater than 10 to 20 ng/ml and Gleason score of 5 to 7.

Adenocarcinoma↗
PubMed1989

Multivariate analysis of prognostic factors in pediatric and adult thalamic and brainstem tumors.

A multivariate analysis of prognostic variables was performed on a retrospective review of 136 patients with presumed or histologically proven primary lesions of the thalamus and brainstem treated by combined surgery and post-operative irradiation or by irradiation alone from January 1950 through December 1983. Overall survival for all patients at 5 and 10 years was 34.4 and 27.8%, respectively. Follow-up of 33 living patients ranged from 3 to 22 years. Prognostic variables analyzed by univariate analysis and found to be of significance (p less than 0.05) were race, duration of symptoms, extent of surgery (i.e. subtotal excision), and dose of irradiation. Further evaluation by Cox regression analysis revealed these same factors to be of prognostic significance (p less than 0.05). It is of importance to note that age and tumor site were not significant prognostic variables in the multivariate analysis but were significant by univariate analysis. The 5-year overall survival for patients with thalamic tumors was 59.5 and 20.9% for children and adults, respectively (p = 0.006). The 5-year overall survival for patients with pontine lesions was 46.6 and 16.0% for adults and children, respectively (p = 0.01). Only one patient was known to have expired due to a complication of therapy. Neurologic deficits and functional ability was normal or mild in 57.6% of the surviving patients.

Adolescent↗
PubMedDate not supplied

Tumour markers CEA and CA 15-3 as Prognostic factors in breast cancer--univariate and multivariate analysis.

Tumour markers are putative prognostic indicators for patients with breast cancer, but have not been elevated independently by multivariate analysis in a large patient number. In 550 patients with breast cancer without known metastases the levels of the serum tumour markers CEA und CA 15-3 were determined preoperatively and during follow-up. The prognostic relevance of these markers for recurrence (n = 128/487) and death of disease (n = 55/550) was evaluated in relation to established prognostic factors. In univariate analysis tumour size, lymph nodes, histological grading, age, hormone receptors, preoperative value of CEA (cut-off 2 ng/mL) and CA 15-3 (cut-off 25 U/mL) and their decrease of more than 33% within seven months after operation were significant for relapse. The results for death of disease were similar except for age. In multivariate analysis tumour size, lymph nodes and decrease of CEA > 33% (p < 0.001) were independent prognostic factors for recurrence. For overall survival tumour size, lymph nodes, histological grading and preoperative levels of CEA > or = 2 ng/mL (p = 0.038) and of CA 15-3 > or = 25 U/mL (p = 0.007) were independent prognostic factors. Pre- and postoperative values of the tumour markers CEA und CA 15-3 are strong independent prognostic factors for relapse and survival in breast cancer patients.

Analysis of Variance↗
PubMed2000

Late rectal toxicity after conformal radiotherapy of prostate cancer (I): multivariate analysis and dose-response.

PURPOSE: The purpose of this paper is to use the outcome of a dose escalation protocol for three-dimensional conformal radiation therapy (3D-CRT) of prostate cancer to study the dose-response for late rectal toxicity and to identify anatomic, dosimetric, and clinical factors that correlate with late rectal bleeding in multivariate analysis. METHODS AND MATERIALS: Seven hundred forty-three patients with T1c-T3 prostate cancer were treated with 3D-CRT with prescribed doses of 64.8 to 81.0 Gy. The 5-year actuarial rate of late rectal toxicity was assessed using Kaplan-Meier statistics. A retrospective dosimetric analysis was performed for patients treated to 70.2 Gy (52 patients) or 75.6 Gy (119 patients) who either exhibited late rectal bleeding (RTOG Grade 2/3) within 30 months after treatment (i.e., 70.2 Gy-13 patients, 75. 6 Gy-36 patients) or were nonbleeding for at least 30 months (i.e., 70.2 Gy-39 patients, 75.6 Gy-83 patients). Univariate and multivariate logistic regression was performed to correlate late rectal bleeding with several anatomic, dosimetric, and clinical variables. RESULTS: A dose response for >/= Grade 2 late rectal toxicity was observed. By multivariate analysis, the following factors were significantly correlated with >/= Grade 2 late rectal bleeding for patients prescribed 70.2 Gy: 1) enclosure of the outer rectal contour by the 50% isodose on the isocenter slice (i.e., Iso50) (p < 0.02), and 2) smaller anatomically defined rectal wall volume (p < 0.05). After 75.6 Gy, the following factors were significant: 1) smaller anatomically defined rectal wall volume (p < 0.01), 2) higher rectal D(max) (p < 0.01), 3) enclosure of rectal contour by Iso50 (p < 0.01), 4) patient age (p = 0.02), and 5) history of diabetes mellitus (p = 0.04). In addition to these five factors, acute rectal toxicity was also significantly correlated (p = 0.05) with late rectal bleeding when patients from both dose groups were combined in multivariate analysis. CONCLUSION: A multivariate logistic regression model is presented which describes the probability of developing late rectal bleeding after conformal irradiation of prostate cancer. Late rectal bleeding correlated with factors which may indicate that a greater fractional volume of rectal wall was exposed to high dose, such as smaller rectal wall volume, inclusion of the rectum within the 50% isodose on the isocenter slice, and higher rectal D(max).

Analysis of Variance↗
PubMed2005

Survival and recurrences after hepatic resection or radiofrequency for hepatocellular carcinoma in cirrhotic patients: a multivariate analysis.

Hepatic resection is still considered the treatment of choice for hepatocellular carcinoma in patients with liver cirrhosis. Radiofrequency ablation is a new emerging modality. The aim of this study was to compare two homogeneous groups of patients who underwent either surgical resection or laparoscopic radiofrequency, analyzing the factors predicting survival and intrahepatic recurrences with use of a multivariate analysis. From February 1997 to April 2003, 98 patients were enrolled in this prospective study. Inclusion criteria were a single nodule of less than 5 cm, Child A-B class of liver function, and no previous treatment: 40 patients were in the surgical group and 58 patients were in the radiofrequency group. The two groups were homogeneous as far as preoperative characteristics were concerned. Operative mortality was zero, and the rates of operative morbidity were similar. Actuarial survival at 4 years was not significantly different (61% after resection and 45% after radiofrequency). There was a significant higher incidence of intrahepatic recurrences after radiofrequency than after resection (53% versus 30%; P=0.018). This was mainly due to local recurrences, whereas those appearing in other liver segments were similar in both groups. A multivariate analysis showed that the significant factors predictive of an intrahepatic recurrence were the level of alpha-fetoprotein, the etiology of cirrhosis, and the type of the treatment. On the other hand, multivariate analysis of the survival showed that only the level of alpha-fetoprotein was an independent predictor of survival. The results of our study showed a significant lower incidence of intrahepatic recurrences after resection compared with after radiofrequency. This seems not to significantly influence the overall survival, probably because of a prompt and effective treatment of the recurrences themselves.

Aged↗
PubMed1997

Preoperative carcinoembryonic antigen predicts outcomes in node-negative colon cancer patients: a multivariate analysis of 572 patients.

BACKGROUND: Although prospective trials have demonstrated that postoperative chemotherapy for node-positive colon cancer patients provides survival benefit, no improvement in survival has been documented for node-negative colon cancer patients. There are, however, a subset of node-negative patients that go on to die of their disease. We hypothesize that this subset of node-negative patients may benefit from postoperative chemotherapy. We analyzed a large cohort of node-negative colon cancer patients from a single institution to determine prognostic factors that predict which patients with node-negative colon cancer might experience recurrence and can benefit from postoperative chemotherapy. STUDY DESIGN: A review of the prospective database for colorectal cancer at Memorial Sloan-Kettering Cancer Center (MSKCC) between 1985 and 1993 identified 572 patients who underwent curative resection for node-negative colon cancer (T(1,2,3,4)N0M0). Demographic, serum, and pathologic factors were analyzed for prognostic significance. Survival was calculated by the method of Kaplan-Meier and compared by log rank test. Multivariate analysis was calculated by the Cox proportional hazard model. RESULTS: Median follow-up was 35 months. Factors predictive of survival by univariate analysis include tumor stage, overall stage, and preoperative serum carcinoembryonic antigen (CEA) elevation. By multivariate analysis, overall stage and preoperative serum CEA level predicted survival. CONCLUSIONS: Routine histologic and demographic factors do not predict outcome in node-negative colon cancer patients. Preoperative CEA and overall stage predict survival by multivariate analysis. Preoperative CEA elevation in node-negative patients identifies a group of patients that has a poor prognosis and defines a subset of patients who may benefit from postoperative chemotherapy.

Adult↗