The importance of measuring plasma carcinoembryonic antigen in small-cell anaplastic carcinoma

The clinical value of the serum biomarker carcinoembryonic antigen (CEA) was evaluated prospectively in 118 patients with small cell lung cancer (SCLC) entered chemotherapy protocol between 1986 and 1992. Five quantitative categories were determined: less than 2.5 ng/ml and 2.6-5.0 ng/ml (the standard normal), 5.1-20.0 ng/ml, 20.1-100 ng/ml and greater than 100 ng/ml. 70% of patients had levels less than 5 ng/ml and only 19% had levels greater than 20 ng/ml. There was no clearcut relationship of plasma CEA level to stage of disease, in which 61% of patients with extensive disease (59 patients) had levels less than 5 ng/ml and 22% of patients with limited disease (59 patients) had levels greater than 5 ng/ml. There was a modest relationship of CEA levels to presence of metastases, in that 50% of patients with metastases had levels greater than 20 ng/ml. The average survival for the pathologic and normal category was almost similar, ranging from 13.27 to 16.81 months. The correlation between disease extent and survival was more sensitive for lactate dehydrogenase (LDH) than for CEA. So CEA as a tumor marker for SCLC must be applied in conjunction with other biomarkers, particularly LDH and neuron specific enolase (NSE) and is meaningful in only a small proportion of patients.     

Carcinoembryonic antigen (CEA) assays in obstructive colorectal cancer

Four of 40 patients with resectable colon or rectal cancer had tumors causing acute large bowel obstruction with colonic dilatation; all 4 patients had preoperative CEA titers above 10 ng/ml with a mean of 28 ng/ml. Thirty-six cancer patients without acute colon obstruction had a mean CEA titer of 4.5 ng/ml; only 6 of 36 patients had circulating CEA titers 10 ng/ml or greater. This suggested that pre-treatment CEA titers in patients with obstructing cancer are unusually high. Multiple CEA assays were performed on two of the 4 patients with colonic obstruction before and after bowel decompressive procedures and prior to their definitive treatment. Relief of obstruction alone produces marked reduction in circulating CEA; this suggested that not only the extent of disease but also the pathophysiological changes associated with obstruction influenced circulating CEA levels.     

Video-imaging and treatment presentation: medico-legal implications and patient perception

Peripheral blood leukocyte alkaline phosphatase (LAP) scores and CA15-3, CA125, and CEA levels in plasma were measured in 57 patients with metastatic breast, ovarian, and colorectal cancer, respectively, and in 79 patients with the same types of nonmetastatic cancer. The mean LAP scores of the metastatic cancer patients (261, 272 and 275 for breast, ovary and colon, respectively) were significantly higher than those of the nonmetastatic cancer group (70, 68 and 57, respectively). There was no overlap between the 95% confidence intervals of the two groups (i.e., metastatic versus nonmetastatic), and no patient known to be metastatic had a LAP score within the normal range. The mean levels of other markers in the metastatic patients (CA15-3, 63.4 mu/ml; CA125, 104.8 mu/ml; and CEA, 51.8 ng/ml for metastatic breast, ovarian, and colon cancer, respectively) were also higher than in the nonmetastatic patients (CA15-3, 24 mu/ml; CA125, 25.3 mu/ml; and CEA, 5.8 ng/ml for nonmetastatic breast, ovarian, and colon cancer, respectively). However, the 95% confidence intervals of the nonmetastatic and the metastatic patients overlapped so that there were false-negatives and/or false-positives when the other markers were used. We therefore conclude that the addition of the LAP score to conventional cancer markers could be helpful for the diagnosis of recurrence and follow-up of cancer patients and suggest that our results be confirmed by further studies on a larger series of patients.     

Bacillus Calmette-Guérin potentiates monocyte responses to lipopolysaccharide-induced tumor necrosis factor and interleukin-1, but not interleukin-6 in bladder cancer patients

During the past decade, particular attention has been focused on treatment of bladder cancer patients with the bacterial agent bacillus Calmette-Guérin (BCG). In these studies, bladder cancer patients were instilled with BCG (75 mg/50 ml) once per week for 6 weeks, 1-2 weeks following trans-urethral resection of the bladder. Cystoscopy was performed after 6 weeks and, unless tumor progression was present, monthly treatments were given for 1 year. Blood was drawn 2 h after the last instillation, and monocytes were isolated (5 x 10(6) cells/ml) and treated, or not, with lipopolysaccharide (LPS) (20 microgram/ml) for tumor necrosis factor alpha (TNF alpha), interleukin-1 alpha (IL-1 alpha) and interleukin-6 (IL-6) release. The levels of monokines were determined by a monokine-specific enzyme-linked immunosorbent assay. Our results clearly show that, after 18 h incubation, macrophages from BCG-treated bladder cancer patients produced from 2.8- to 1.9-fold and from 2.0- to 1.3-fold greater amounts of TNF alpha and IL-1 alpha respectively, compared to macrophages from healthy controls, 5-fold higher than bladder cancer patients not treated with BCG. IL-6 was not affected. In another set of experiments macrophages (5 x 10(6) cells/ml) from healthy subjects were pretreated, or not, with BCG (100 micrograms/ml) overnight and treated, or not, with LPS 20 microgram/ml alone and in combination with interleukin-1 receptor antagonist (IL-1ra) 250 ng/ml. Macrophages treated with BCG had a strong stimulatory effect on IL-1 alpha release (9.45 ng/ml) while LPS was less effective (3.59 ng/ml). The combination of BCG plus LPS produced an additive effect on IL-1 alpha release (13.71 ng/ml) compared to the effect of the compound alone. The addition of IL-1ra (250 ng/ml) to BCG was not effective, while when IL-1ra was added to BCG plus LPS only a partial inhibition of IL-1 alpha release was found (9.83 ng/ml), compared to BCG plus LPS without IL-1ra (13.71 ng/ml). These effects seem to be related to the inhibition of IL-1 alpha stimulated with LPS, but not BCG. The priming effect of BCG exerted on LPS-stimulated monocyte production of TNF alpha and IL-1 alpha from bladder cancer patients led us to study the possible modulation of fibrinogen and C-reactive protein in the serum of BCG-treated cancer patients. The plasma levels of fibrinogen and C-reactive protein were higher (approximately twice) in BCG-treated patients compared to values obtained in untreated patients or healthy controls. We conclude that the beneficial immunotherapeutic effects of BCG in bladder cancer patients are related to its capacity to prime macrophages to enhance the release of TNF alpha and IL-1 alpha, but not IL-6 in response to physiological secondary stimuli, or through the direct stimulation of BCG on IL-1 alpha or TNF alpha, which are directly involved in the killing of cancer cells.(ABSTRACT TRUNCATED AT 400 WORDS)     

Prognostic use of preoperative and immediate postoperative carcinoembryonic antigen determinations in colonic cancer

Preoperative and postoperative carcinoembryonic antigen (CEA) concentrations are useful in predicting the likelihood of recurrence in patients undergoing curative operations for cancer of the colon. The incidence of recurrence was studied in three groups of patients followed for 6 to 18 months after such an operation: 36 patients (group 1) had preoperative and postoperative plasma CEA concentrations less than 2.5 ng/ml; 11 patients (group 2) had a preoperative CEA value above but a postoperative value below 2.5 ng/ml; and 11 patients (group 3) had preoperative and postoperative concentrations greater than 2.5 ng/ml. Cumulative rates of recurrence at 6, 12 and 18 months after operation were as follows: 0, 3 and 17% in group 1; 0, 9 and 21% in group 2; and 27, 79 and 79% in group 3. Statistically there was no difference in the recurrence rate between groups 1 and 2, but the rate was significantly lower in group 2 compared with group 3. The data show that patients in whom the immediate postoperative CEA concentration returns to normal have a much lower incidence of recurrent cancer of the colon than patients whose CEA concentration remains elevated.     

Quantitative analysis of CEA expression in colorectal adenocarcinoma and serum: lack of correlation

Tissues and sera from 110 patients diagnosed with colorectal primary carcinoma, 20 patients with benign colorectal diseases and 31 healthy donors were subjected to quantitative CEA analysis. Multiple samples from tumor lesions and autologous histologically normal mucosa (10 cm from the tumor) were obtained at the time of surgery (cancer patients) or endoscopy (benign patients and healthy volunteers). CEA content was measured in protein extracts obtained from these tissues using a quantitative RIA method. A limit of normality for CEA content was established as 300 ng/mg of protein. When this was taken as cut-off, 104 of 110 (94.5%) tumor lesions and 51 of 110 (46.4%) autologous histologically normal colonic mucosa from cancer patients had elevated CEA levels. No correlation with stage of disease was found, while a correlation was observed with degree of tumor differentiation. A statistically significant difference between CEA content in tumor lesions and in histologically normal mucosa from cancer patients was observed (p = -0.001). Moreover, CEA content was statistically higher in the normal mucosa from cancer patients than in that from healthy donors (p = 0.005). CEA content in tissue specimens from benign lesions differed significantly from that in tissue from healthy donors (p = 0.005) and in carcinoma lesions (p < 0.001). The highest CEA content was observed in benign lesions with severe dysplasia. No statistical correlation between CEA content in carcinoma tissues and serum CEA levels (r = 0.195, p = .13) was found. Therefore, in considering diagnosis or therapy with anti-CEA MAbs for colorectal-carcinoma patients, or potential therapies with anti-CEA recombinant vaccines, serum CEA levels should not be taken as indicating CEA expression in tumor lesions.     

Surgical treatment methods in combined nasal deformities

Different classes of antibodies to polycyclic aromatic hydrocarbons (PAH) and immunogenicity of carcinoembryonic antigen (CEA) were investigated in breast cancer patients versus course. Prior to treatment, the levels of IgM, IgG and IgA antibodies to PAHs were 5.9, 7.7 and 32.5%, respectively. In cases of tumor regression, they were 13.8, 25.0 and 19.2%, while in those of progression-2.4, 2.8 and 12.8%, respectively. The immunogenicity of CEA, i.e. the power of forming circulating immune complexes, appeared to be higher in cases of tumor regression than in those of progression. CEA concentration in immune complexes exceeding 5ng/ml was observed in 24.0 and 15.1% of cases, respectively, while in whole serum, the concentration in excess of 10 ng/ml was in 8.0 and 42.4%, respectively.     

Reply to Burghes

We investigated whether the concentration of circulating transforming growth factor beta (TGFbeta) yields diagnostic value in breast cancer. Blood was collected from twenty stage I and II breast cancer patients both prior to treatment and after surgical excision of the tumour. Both latent and active TGFbeta were quantified directly in the blood plasma using a bioassay. The average plasma TGFbeta level in breast cancer patients was 20.8 +/- 8.5 ng/ml (n=20; mean +/- SD), which was not different from normal controls. Elevated plasma TGFbeta levels (> average control +/- 2SD) were found in 5% (1/20) of the controls and in 25% (5/20) of the patients. Correlation was not found between plasma TGFbeta level and tumour type nor with tumour stage. Following surgical excision of the tumour, plasma TGFbeta levels were not significantly altered. Thus, our data show that plasma TGFbeta levels do not reveal diagnostic value for early stage breast cancer. Determination of the pretreatment plasma TGFbeta value of the individual patient might, however, still be meaningful since it appears to be predictive for its normal tissue reaction following cancer therapy.     

Soluble intercellular adhesion molecule-1 in patients with lung cancer and benign lung diseases

Intercellular adhesion molecule-1 (ICAM-1) expression correlates with tumour progression in patients with malignant melanoma or renal cell carcinoma. To assess the value of soluble ICAM-1 (sICAM-1) for lung cancer patients, sICAM-1 was determined by means of an enzyme-linked immunosorbent assay. Sera from 147 patients with lung cancer, from 75 patients with benign lung diseases and from 108 healthy adults were investigated for sICAM-1 expression. Significant differences in sICAM-1 levels were detected in lung cancer patients (387 +/- 176 ng/ml) and patients with benign lung diseases (365 +/- 110 ng/ml) compared to the group of healthy adults (310 +/- 90 ng/ml). There was no difference in sICAM-1 level among the subtypes of lung cancer. Advanced tumour stages and patients with progressive disease tended to be associated with higher sICAM-1 levels, the site of metastasis being relevant for the level attained. Patients with liver metastasis had the highest sICAM-1 levels (547 +/- 295 ng/ml) compared to patients with cerebral metastasis (317.8 +/- 92.2 ng/ml). An increase of sICAM-1 expression during the progression of the disease coincided with a poorer survival prognosis for the patients compared to patients with stable or falling sICAM-1 levels.     

The evaluation of carcinoembryonic antigen determination in the pericardial fluid in the diagnosis of the cause of pericarditis

Pericardial fluid CEA level was measured with radioimmunoassay in 19 patients with large pericardial effusion of unknown origin. In 11 patients malignancy was diagnosed. In all of these patients pericardial fluid CEA levels were above 7 ng/ml (mean value 52.6 +/- 42.6 ng/ml). In 8 patients the etiology of pericarditis was non-malignant. In all of them pericardial fluid CEA levels were below 7 ng/ml (mean value 2.2 +/- 1.6 ng/ml). In 9 patients with malignant pericarditis serum CEA levels were also determined: they were found to be lower than pericardial fluid CEA values in 6 patients. It was concluded that pericardial fluid CEA elevation is a reliable criteria of neoplastic pericardial involvement.     

Effect of a poly(ADP-ribose) polymerase inhibitor on DNA breakage and cytotoxicity induced by hydrogen peroxide and gamma-radiation

Thirty-six outpatients aged 20 to 51 with RDC primary major depressive disorder (MDD) completed a 5-week trial of desipramine following a week of single-blind placebo. Five had a past history of hypomanic disorder. For all but one patient, daily dosage at bedtime was constant for the final 4 weeks, with a mean (S.D.) of 168.1 (46.5) mg. Plasma samples drawn at the three final weekly visits were assayed by high-performance liquid chromatography for 2-hydroxydesipramine (2-OH-DMI) and desipramine. Mean (S.D.) plasma levels were 59.8 (30.0) ng/ml for 2-OH-DMI and 142.9 (138.6) ng/ml for desipramine. Thirteen patients (36%) had a final 17-item Hamilton depression rating < and = 6 and were classified as responders. According to receiver operating characteristics analysis, patients with plasma 2-OH-DMI levels > and = 58 and < 92 ng/ml had a greater likelihood of responding than those with lower or higher levels (p = 0.005, Fisher's exact test), while patients with plasma desipramine levels > and = 64 ng/ml were more likely to respond than those with lower levels (p = 0.032, Fisher's exact test). Results using an alternate response criterion were similar. These findings suggest that in desipramine-treated outpatients with primary MDD the relationship between therapeutic response and plasma levels is curvilinear for 2-OH-DMI and linear for desipramine.     

Birthmarks to worry about. Cutaneous markers of dysraphism

CYFRA 21-1 is a fragment of cytokeratin 19 (CK 19). Four patients with large intrahepatic (or peripheral) cholangiocarcinoma (CC) and high serum levels of CYFRA 21-1 (normal, < or = 2 ng/ml) are reported. CYFRA 21-1 levels exceeded 9 ng/ml in all 4 patients. Carcinoembryonic antigen (CEA), was high in 1 (CEA; normal range, < or = 5.0 ng/ml) and carbohydrate antigen 19-9 (CA 19-9) was high in 3 (CA19-9; normal range, < or = 36 U/ml). We also measured serum levels of CYFRA 21-1 in 13 patients with hepatocellular carcinoma (HCC) more than 5 cm in diameter. Levels of CYFRA 21-1 exceeded 2 ng/ml in 9 of the HCC patients and were higher than 9 ng/ml in 2 of the HCC patients. Levels of alpha fetoprotein (AFP) and/or protein induced by vitamin K absence or antagonist II (PIVKA II) were elevated in all HCC patients (AFP, PIVKA II, respectively; normal range, < or = 10.0 ng/ml and < or = 0.1 AU/ml) CYFRA 21-1 levels were measured twice or three times during the clinical course in 2 CC patients and in 6 HCC patients, and increased gradually with tumor growth in the 2 CC patients and in 3 of the 6 HCC patients. Marked increases in serum CYFRA 21-1 levels in patients with large liver cancers, particularly in those with normal levels of AFP and PIVKA II, would suggest the existence of intrahepatic CC rather than HCC.     

Screening for prostate cancer

In an attempt to detect prostate cancer when the tumor is still confined to the prostate, a screening program was established. We studied the efficacy of digital rectal examination (DRE) and serum prostate-specific antigen (PSA) in the early detection of prostate cancer. One thousand men aged 50-75 years underwent DRE and serum PSA determination. Transrectal ultrasound-guided biopsies were obtained in each case of a suspicious DRE. Six systematic biopsies were performed if the PSA level was > 10 ng/ml, even if DRE and transrectal ultrasonography revealed no areas suspicious of cancer. A suspicious DRE was noted in 11.5% of the subjects; 16% had elevated levels of serum PSA (> 4 ng/ml) and 3.9% had serum PSA > 10 ng/ml. Biopsies were obtained from 90 patients, of which 31 were positive for prostate cancer. The cancer detection rate was 2.2% for DRE, 2.0% for PSA > 10 ng/ml, and 3.1% for the two methods combined. Clinical staging revealed that in 29 of the 31 patients with prostate cancer, the tumor was confined to the prostate: Stage A in 9 cases and stage B in 20 cases. Only two patients had clinically advanced cancer, and 22 patients underwent radical prostatectomy. Pathological examination disclosed biologically significant tumors in 91% of the cases in terms of tumor volume and grade. Although there is little evidence that screening will result in the reduction of the disease-specific mortality rate, early detection of prostate cancer by DRE, serum PSA, and transrectal ultrasound should be encouraged.     

An analysis of preliminary results in screening for prostatic cancer]

450 males aged over 50 years free of urological symptoms were screened for prostatic cancer using three techniques; finger rectal examination (FRE), transrectal ultrasound investigation (TUI), assay for prostatic specific antigen in the serum (SPSA). SPSA quantities under 4 ng/ml, 4-10 ng/ml, 10-20 ng/ml, over 20 ng/ml were registered in 206(45.8%), 135(30%), 69(15.4%) and 40(8.8%) patients, respectively. Detectability of prostatic cancer increases by 33,37.9, 45.5, 69.2% due to TUI, FRE, TUI + FRE, all the three methods, respectively. Prostatic biopsy was needed in 102 (22.7%) cases. From the 450 examinees, prostatic cancer was diagnosed in 25 (5.6%). SPSA was high in all of them, higher than 10 ng/ml in 92%. 20 (80%) of 25 patients with cancer had early stages of the disease (TI-2). The study is going on.     

Comparative study of carbohydrate antigen 195 and carcinoembryonic antigen for the diagnosis of pancreatic carcinoma

The expression CA195 in serum, defined by monoclonal antibody CC3C195 (IgM), was studied in 67 patients with pancreatic cancer and in 138 patients with biliary or pancreatic benign disease. The results were compared with carcinoembryonic antigen (CEA) expression. The overall sensitivity of the CA195 assay (> 12 U/ml) was higher than that for CEA (89.5% vs. 53.7%) (p < 0.001). Sensitivity was increased to 92.5% with the simultaneous use of the two antigens, but the difference was statistically significant only with CEA (p < 0.001). The specificity of CA195 calculated from all patients with benign diseases was lower than that of CEA (73.1% vs. 89.8%). However, using a cutoff value of 100 U/ml for CA195, the specificity of this antigen (82%) was higher than that of CEA. These results demonstrate that marked elevations of tumor antigen CA195 are relatively specific for pancreatic carcinoma, and that this antigen is superior to CEA for diagnosing pancreatic cancer by virtue of its higher sensitivity.     

MR arthrography of the shoulder: rethinking traditional imaging procedures to meet the technical requirements of MR imaging guidance

Increased serum levels of mucin-associated antigen have been previously demonstrated in patients with cystic fibrosis (CF) and interstitial pneumonia, and in lung-transplant recipients. The present study assessed the serum airway mucin levels in patients with acute respiratory distress syndrome (ARDS). An enzyme-linked immunosorbent assay (ELISA) method with a human-airway-mucin-specific monoclonal antibody (17Q2) was used to measure serum mucin levels in normal subjects, chronic smokers, patients with chronic bronchitis and other pulmonary diseases, patients with acute cardiogenic lung edema, and patients with ARDS. The serum mucin levels measured 9.9 +/- 0.8 ng/ml (mean +/- SEM, n = 59) in normal subjects, 12.7 +/- 1.6 ng/ml (n = 29) in chronic smokers, 21.8 +/- 1.9 ng/ml (n = 28) in patients with chronic bronchitis and other pulmonary diseases, 9.0 +/- 3.1 ng/ml (n = 5) in patients with acute cardiogenic lung edema. The serum mucin level was 53.8 +/- 6.6 ng/ml (n = 13) in patients with ARDS (p < 0.05, as compared with the four other groups). Serial measurements of serum mucin levels were obtained in patients with ARDS. Statistical analysis showed an inverse correlation of serial measurements of serum mucin with static respiratory-system compliance (p = 0.021), an inverse correlation of sequential serum mucin levels and log(Pa(O2)/Fl(O2)) (p = 0.016), and a positive correlation of sequential serum mucin levels and lung injury score (LIS) (p = 0.019). Gel-filtration analysis showed that mucin-associated antigens in ARDS sera were polydispersed and smaller than the antigens in normal sera. This study indicates that an increasing amount of degraded mucin occurs in patients with ARDS.     

Physical and genetic localization of the gene encoding the AP-2 transcription factor to mouse chromosome 13

This retrospective study assessed 46 patients with advanced ovarian cancer who had elevated preoperative serum CA 125 (>35 U/ml) and who had another serum CA 125 assay 6-9 days after surgery. Preoperative CA 125 levels were similar in patients with residual disease below 20 mm and in those with larger residuum. The postoperative decline of serum CA 125 was significantly higher in patients with small residual disease at any preoperative serum CA 125 value. By taking 60% as the cutoff of CA 125 decline, the diagnostic accuracy of this parameter in discriminating between patients with residual disease below or above 20 mm improved progressively when we considered patients with increasing preoperative antigen values. However, even in the subset of patients with preoperative serum CA 125 above 400 U/ml, 2 of the 20 patients with less than 20 mm residual disease had a percentage reduction of antigen levels lower than 60%, whereas 5 of the 10 patients with larger than 20 mm residuum had a CA 125 decline higher than 60%. Therefore, we believe that the perioperative changes of CA 125 levels have a limited clinical relevance in the management of patients with ovarian cancer.     

Stage T1-2 prostate cancer: a multivariate analysis of factors affecting biochemical and clinical failures after radical prostatectomy

PURPOSE: Prostate-specific antigen (PSA) is extensively used in case selection and outcome evaluation after treatment of clinically localized prostate cancer. Careful case selection can have a profound impact on pathologic findings and ultimate outcome. In addition, salvage treatment is frequently initiated at the time of biochemical relapse rather than clinical recurrence. Consequently, patterns of failure can be significantly altered compared to previous times when PSA was not available. To better understand the impact of PSA on pathologic findings, outcome, and salvage treatment, we reviewed our experience in the PSA era with clinical Stage T1-2 prostate cancer treated with radical prostatectomy. METHODS AND MATERIALS: Between 1987 and 1993, 423 cases could be identified with clinical Stage T1-2 prostate cancer treated with radical prostatectomy. The distribution of cases by pretreatment PSA levels was as follows: < or = 4 ng/ml (18%), 4-10 ng/ml (42%), 10-20 ng/ml (21%), > 20 ng/ml (14%), and unknown (5%). The median pretreatment PSA level for the entire group was 8.0 ng/ml. Sixteen patients received adjuvant or neoadjuvant androgen suppression and 13 received postoperative radiotherapy. Only 31 patients (7%) had pathologically positive pelvic lymph nodes. The overall margin involvement rate was 46%. Fifty-three percent of patients had surgical Gleason scores > or = 7, and 65% had extracapsular extension. The median follow-up time was 41 months. RESULTS: The projected overall survival at 7 years after surgery was 90%. The 5-year clinical relapse-free survival rate was 84%. At 5 years, the local control and distant failure rates were 92% and 91%, respectively. Biochemical relapse was defined as a detectable or rising PSA level after prostatectomy. The 5-year biochemical relapse-free survival (bRFS) rate was 59%. The 5-year RFS was 88% in patients with preoperative PSA levels < or = 4, 62% for 4-10, 48% for 10-20, and 31% for > 20. Combining the two independent preoperative variables, iPSA and biopsy GS (bGS), two risks groups were defined: low risk [initial PSA (iPSA) levels < or = 10.0 and bGS < or = 6] and high risk (iPSA levels > 10.0 ng/ml or bGS > or = 7). The 5-year bRFS rate for the low-risk cases was 81% vs. 40% for high-risk cases (p < 0.001). On multivariate analysis, three factors independently predicted biochemical relapse: iPSA levels (p = 0.005), Gleason score from the surgical specimen (sGS) (p = 0.002), and positive surgical margins (p < or = 0.001). The 5-year bRFS rates for margin positive vs. margin negative patients were 37% vs. 78%, respectively. The 5-year bRFS rates for GS > or = 7 vs. GS > or = 6 were 42% vs. 80%, respectively. All clinical relapses were accompanied by a rise in PSA. In patients who manifested biochemical failure followed by a clinical failure, the median interval between the PSA rise and clinical failure was 19 months (range 7-71). Margin involvement was the only independent predictor of local failure (p = 0.019). The 5-year local failure-free survival for negative margin cases was 96% vs. 87% for positive margin cases (p = 0.012). Lymph node (LN) involvement and high-risk group were the two independent predictors of distant failure. The 5-year distant failure-free survival for negative LN cases was 94% vs. 67% for positive LN cases (p < 0.001). The 5-year distant failure-free survival for low-risk cases was 97% vs. 85% for high-risk cases (p = 0.005). For the 124 patients failing biochemically, 85 were observed and 39 were treated either with radiation or androgen deprivation. With a median follow-up of 32 months, the clinical disease relapse-free survival was 79% for the treated patients vs. only 32% for the patients observed (p < 0.001). CONCLUSION: Pretreatment PSA is the most potent clinical factor independently predicting biochemical relapse, thereby allowing markedly better case selection. Achieving negative margins, even in relatively advanced disease, provides excellent lon