Dangers of using "optimal" cutpoints in the evaluation of prognostic factors

Lack of prognostic significance of epidermal growth factor receptor and the oncoprotein p185HER-2 in patients with systemically untreated non-small-cell lung cancer: an immunohistochemical study on cryosections

The prognostic role of the epidermal growth factor receptor (EGFR) and the related receptor p185HER-2 in lung cancer is as yet undefined. We investigated the immunohistochemical expression of EGFR (monoclonal antibody R1; Amersham) and p185HER-2 (polyclonal antibody A485; Dako) in cryosections. A total of 186 unselected and systemically untreated patients with non-small-cell lung cancer (NSCLC) diagnosed and treated at Odense University Hospital, Denmark, were included. Median follow-up period was 66 months. EGFR and p185HER-2 was highly expressed in 55% and 26% of cases respectively. Expression of EGFR was independent of p185HER-2 expression. The expression of EGFR was higher in squamous cell carcinomas whereas the level of p185HER-2 staining was higher in adenocarcinomas. Expression of either or both receptors was not correlated with age, histological grading, stage and prognosis. We conclude that immunohistochemical detection of these growth factor receptors failed to demonstrate a prognostic significance in patients operated on for NSCLC.

Intensive treatment of patients age 60 years and older with de novo acute myeloid leukemia: analysis of prognostic factors

BACKGROUND

This study aimed to define pre-treatment parameters with prognostic significance in elderly patients with de novo acute myeloid leukemia (AML) who were treated with aggressive regimens.

METHODS

We analyzed, retrospectively, the clinical and laboratory features of 159 consecutive patients age >60 years with AML. Ninety-two patients presenting as de novo AML were considered suitable for aggressive chemotherapy according to inclusion criteria not different from those commonly used for younger adults. They belonged to all of the French-American-British classification types except M3, and their median age was 67 years (range: 60-79). Antileukemic treatment consisted of 1 of 3 sequential protocols adopted at the S. Eugenio University Hospital of Rome between 1987 and 1993. The three therapeutic groups were similar in number and presenting characteristics. In addition to arabinosylcytosine, induction schedules included mitoxantrone (Groups I and II) or daunorubicin (Group III), and etoposide (Groups I and III). Once in complete remission (CR), patients were consolidated with two other courses of chemotherapy using reduced dosages of the same drugs given during induction.

RESULTS

Induction treatment achieved a 52.2% CR rate, with median remission duration and event free survival (EFS) of 35 and 27 weeks, respectively. Because no significant differences between the results of the three therapeutic groups were observed, all cases were pooled to evaluate the prognostic factors. In univariate analysis, the only presenting characteristic significantly associated with failure of induction treatment was age >67 years (P=0.007). Factors associated with an increased likelihood of shorter remission duration were CD7 expression on leukemic cells (P=0.007) and an abnormal karyotype (P=0.010; those predicting shorter EFS were a chromosomal status other than normal (P=0.002) and detection of CD14 antigen (P=0.008). Logistic regression results identified age and CD14 expression as the variables with independent prognostic impact on CR achievement. In a stepwise proportional hazards general linear model, CD7 and karyotype retained their predictive value regarding remission duration, whereas the karyotypic pattern at diagnosis and CD14 antigen expression were the most important determinants of EFS, with age showing a borderline statistical value. A simple "risk factor score" was developed that would allow for stratification of patients into prognostic groups.

CONCLUSIONS

Cytogenetic analysis and immunophenotyping might help to select elderly patients with AML who have little benefit from current therapeutic strategies and with whom new approaches might be experimented.

Lack of prognostic significance of the monoclonal antibody Ki-S1, a novel marker of proliferative activity, in node-negative breast carcinoma

In a series of 205 node-negative breast cancers (NNBC), we determined staining by the novel antibody Ki-S1, a marker of tumor cell proliferation, in order to test its association with other prognostic variables and its prognostic significance. Ki-S1 was determined in routinely formalin-fixed paraffin-embedded tumor samples. Ki-S1 gave a nuclear staining in the majority of the carcinomas (188 of 205), with percentages of reacting nuclei ranging from 2% to 90% (median value of 7%). In 107 tumors frozen sections were available to also assess the Ki-67 antibody. Among these, 94 had a nuclear staining of cancer cells ranging from 5% to 80% (median value of 7%). In 46 tumors we also determined the MIB-1 antibody. The percentage of MIB-1 nuclear staining ranged from 1% to 50% (median value of 20%). There was no significant relationship between Ki-S1 and the other two cell kinetic markers. Ki-S1 labeling was significantly associated only with tumor size (p = 0.03). With a median follow-up of 6 years, Ki-S1 had no significant prognostic value for either relapse-free survival (RFS) or overall survival (OS) (Ki-S1 as continuous logarithmic variable; p = 0.86 and p = 0.23, respectively). For RFS the following variables had a significant prognostic value: Ki-67 (< or = 10% vs > 10%; p = 0.037); progesterone receptor (PgR) expression (- vs+/++; p = 0.041); tumor size (pT1 vs pT2-3; p = 0.042) and grading (GI vs GII-III; p = 0.047). For OS, tumor size (p = 0.0044), age (continuous variable; p = 0.0060), and Ki-67 (p = 0.043) were significantly prognostic. In multivariate analysis (final model), only tumor size retained a significant and independent prognostic value for RFS (p = 0.0042). For OS, both tumor size (p = 0.0029) and age (< or = 55 years vs > 55 years; p = 0.041) retained significance in the multivariate model. In conclusion, Ki-S1 does not seem to have prognostic relevance in this series of NNBC. Possible hypotheses to explain this observation are discussed.

Relationship between p53 gene abnormalities and other tumour characteristics in breast-cancer prognosis

The prognostic significance of p53 gene abnormalities was investigated in 919 primary breast-cancer patients. p53 expression and tumour-cell proliferation fraction determined by MIB-1 count, p53 exon 5 and 6 mutations and HER-2/neu oncogene amplification were detected by immunohistochemistry, PCR-SSCP and slot-blot hybridization, respectively. Increased MIB-1 count, p53 expression, HER-2/neu oncogene amplification and p53 mutations were detected in 33%, 29%, 10% and 8% of tumours, respectively. Statistically significant associations were observed between p53 expression or MIB-1 count and age below 50 years, high-grade tumours, medullary carcinomas, and absence of hormone receptors. p53 mutations were associated with increased MIB-1 count, HER-2/neu oncogene amplification and absence of hormone receptors, but not with age, tumour size or grade, histological subtype, or the number of axillary nodes involved. After a median follow-up of 66 months, p53 expression was observed to be associated with significant increases in risk of both relapse and death from breast cancer, but not after adjusting for the effect of other parameters. In these analyses, MIB-1 count, and not HER-2/neu oncogene amplification, was an independent predictor of prognosis. In node-negative patients, only p53 exon 5 and 6 mutations and MIB-1 count were associated with a statistically significant increase in risk of death from breast cancer, independent of tumour size and ER concentration. We conclude that tumour-cell proliferation fraction, as measured by MIB-1 count, is the most useful parameter of breast-cancer prognosis, with the exception of ER, tumour size and the number of axillary nodes involved.

An investigation of cut-points for primary breast cancer oestrogen and progesterone receptor assays

Oestrogen and progesterone receptor (ER and PgR) assay values are frequently used in medical decision-making for breast cancer patients. We have proposed statistical standardization of receptor assay values to improve inter-laboratory comparability, and now report the use of standardized log units (SLU) to investigate the effects of ER and PgR cut-points on time to first recurrence outside the breast (DFS). Between 1980 and 1986, there were 678 primary breast cancer patients treated at the Henrietta Banting Breast Centre (HBBC). The effects of ER and PgR cut-points were examined with multivariate analyses considering the variables: age, tumour size, nodal status, weight and adjuvant treatment. We considered receptor assay cut-points ranging from - 1.0 to + 1.0 SLU (ER between 7 and 166 fmol/mg protein; PgR between 7 and 181 fmol/mg protein). PgR was included in the multivariate prognostic models more often than ER, although patients had a better prognosis with both larger ER and PgR values. There was no best cut-point for ER or PgR, and there was strong evidence that ER and PgR should be considered as continuous rather than dichotomous (negative, positive) variables. Patient prognosis should also be more comparable with SLU.

Prospective study of the prognostic significance of epidermal growth factor receptor in primary breast cancer

Bearing in mind the continuing controversy over the prognostic significance of epidermal growth factor receptor (EGF-r) expression, we investigated its clinical significance prospectively in 345 primary breast cancer patients. The prognostic significance of EGF-r expression, as measured by a radioligand binding assay, was determined by Cox's multivariate analysis using EGF-r concentration as a continuous or dichotomous variable. Increased EGF-r expression was detected in 20-32% of tumours, depending on the cut-off in concentration used. EGF-r expression, irrespective of the cut-off, was not associated with tumour size or grade or the number of axillary nodes involved. There was, however, a strong inverse association between EGF-r expression and the absence of hormone receptors. After a median follow-up period of 57 months, multivariate analysis suggested that EGF-r expression was associated with increases in risk for both relapse and death from breast cancer, even after adjusting for oestrogen receptor (ER) concentration, tumour size and the number of axillary nodes involved. Patients with ER-positive tumours, which also expressed EGF-r, had increases in risk for both relapse and death from breast cancer compared with tumours without EGF-r. Expression of EGF-r was not a predictor of poor prognosis in either node-negative or ER-negative subgroups of patients.

Review of survival analyses published in cancer journals

Survival analysis has found widespread applications in medicine in the last 10-15 years. However, there has been no published review of the use and presentation of survival analyses. We have carried out a systematic review of the research papers published between October and December 1991 in five clinical oncology journals. A total of 132 papers were reviewed. We looked at several aspects of study design, data handling, analysis and presentation of the results. We found that almost half of the papers did not give any summary of length of follow-up; that in 62% of papers at least one end point was not clearly defined; and that both logrank and multivariate analyses were frequently reported at most only as P-values [63/84 (75%) and 22/47 (47%) respectively]. Furthermore, although many studies were small, uncertainty of the estimates was rarely indicated [in 13/84 (15%) logrank and 16/47 (34%) multivariate results]. The procedure for categorisation of continuous variables in logrank analyses was explained in only 8/49 (16%) papers. The quality of graphs was felt to be poor in 43/117 (37%) papers which included at least one survival curve. To address some of the presentational inadequacies found in this review we include new suggested guidelines for the presentation of survival analyses in medical journals. These would complement the statistical guidelines recommended by several clinical oncology journals.

Immunohistochemical assay for epidermal growth factor receptor on paraffin-embedded sections: validation against ligand-binding assay and clinical relevance in breast cancer

Epidermal growth factor receptor (EGFR) has been the subject of much research since it was first described as a prognostic factor in breast cancer. The assay methods used and results obtained vary widely between studies. In this study 88 primary breast cancers were assayed for EGFR using a novel immunohistochemical assay performed on paraffin-embedded sections. The monoclonal antibody used was raised against purified, denatured EGFR, reacts with an epitope on the external domain and does not interfere with ligand binding. Twenty-two per cent of the tumours were EGFR positive using this assay. The results obtained were significantly correlated with those obtained by ligand-binding assay (r = 0.621, P = 0.011). The concordance rate was 82% (P < 0.001). The majority of discordant results could be explained by the presence of benign breast tissue and other non-malignant elements which could be seen to express EGFR on the immunohistochemical assay and were excluded from the score for this, but would be incorporated into ligand-binding assay results. The well-established inverse relationship between EGFR (as measured by this assay) and oestrogen receptor (ER) was seen (chi 2 = 24.9, P < 0.0001). In addition, in this exploratory study on a limited tumour set, EGFR was a significant adverse prognostic factor (on univariate but not multivariate analysis) for both relapse-free survival (P = 0.02) and overall survival (P = 0.03) when measured by this immunohistochemical assay, but was not significant when measured by ligand-binding assay.

Immunohistochemical detection of P53 protein is not associated with a poor prognosis in non-small-cell lung cancer

Immunohistochemical detection of the p53 gene product by monoclonal antibodies has been shown to be associated with a poor clinical outcome in carcinomas of the breast and stomach. Because the prognostic relevance of p53 immunostaining in lung cancer is still under debate, we studied the expression pattern and clinical significance of such staining in 73 patients with operable non-small-cell lung cancer. p53 expression was detected on frozen sections with the use of monoclonal antibody p1801, which recognizes both the wild-type and mutant gene product (alkaline phosphatase-anti-alkaline phosphatase method). A tumor was considered p53 positive if more than 1% of the tumor cells were stained. The p53 expression pattern was compared with clinicopathologic parameters, and analysis of follow-up, based on the data of 65 patients, was done by a log rank test (median observation time, 780 days). Nuclear p53 staining was detected in 33 of 73 non-small-cell lung cancers (45.2%). Comparison with clinicopathologic parameters demonstrated that the p53 protein was detected more frequently in younger patients (younger than 50 years, p = 0.014), whereas no correlation was found with sex, tumor differentiation, tumor histologic type, or TNM stage. Surprisingly, follow-up analysis revealed that p53 staining was associated with an increased rate of disease-free survival, especially in patients with early stage tumor disease (p = 0.004) and in male patients (p = 0.023). Counter to previous studies in other solid tumors, immunocytochemical detection of p53 expression does not predict a poor clinical outcome in non-small-cell lung cancer. In early-stage lung cancer it might be associated with an improved disease-free survival, which suggests that the majority of the detected protein inherits the wild-type tumor suppressor function.