Prognostic value of continuous variables in breast cancer and head and neck cancer. Dependence on the cut-off level

Kaplan-Meier Curves, Cox Model, and P-Values Are Not Enough for the Prognostic Evaluation of Tumor Markers: Statistical Suggestions for a More Comprehensive Approach

The assessment of prognostic markers is key to the improvement of therapeutic strategies for cancer patients. Some promising markers may fail to be applied in clinical practice, or some useless markers may be applied, because of misleading results ensuing from inadequate planning of the study and/or from an oversimplified statistical analysis. This commentary illustrates and discusses the main issues involved in planning an effective clinical study and the subsequent statistical analysis for the prognostic evaluation of a cancer marker. Another aim is to extend the most applied statistical models (ie, those using Kaplan-Meier and Cox) to enable the choice of the best-suited methods for study endpoints. Specifically, for tumor-centered endpoints like tumor recurrence, the issue of competing risks is highlighted. For markers measured on a continuous numerical scale, a loss of relevant prognostic information may occur by setting arbitrary cutoffs; thus, the methods to analyze the original scale are explained. Furthermore, because the P-value is not a sufficient criterion to assess the usefulness of a marker in clinical practice, measures for evaluating the ability of the marker to discriminate between "good" and "bad" prognoses are illustrated. Several tumor markers are considered both in human and veterinary medicine. Given the similarity between markers for human breast cancer and canine mammary cancer, an application of the statistical methods discussed within the article to a public dataset from human breast cancer patients is shown.

Cell kinetics and biochemical parameters in breast cancer

The study analyzes biochemical and cell kinetic parameters to characterize solid tumor growth in humans. The concentrations of polyamines, CEA, the thymidine labeling index (T.L.I.) and the mitotic index (M.I.) were determined on fragments of neoplastic tissue from 18 patients with breast carcinoma. Urinary polyamines were evaluated in the same patients.

Two groups of patients were distinguished according to the median value of the T.L.I. In the group with high T.L.I., M.I. and tissue polyamines were significantly higher than in the group with low T.L.I., whereas tissue CEA was lower, though in a not statistically significant way. Urinary polyamines showed no variations between groups.

These preliminary results showed that T.L.I. levels were higher in patients who relapsed during a 4-year follow-up than in patients achieving complete remission and remaining disease free. Results concerning polyamine concentration showed that the tissue polyamine level in breast carcinoma indicated proliferative activity, but this does not seem to be valuable for current prognostic purposes

Quantitative assessment of effective regurgitant orifice: impact on risk stratification, and cut-off for severe and torrential tricuspid regurgitation grade

Aims 

Asses the added value of quantitative evaluation of tricuspid regurgitation (TR), the proper cut-off value for severe TR and ‘torrential TR’ based on outcome data. The added value of quantitative evaluation of TR, and the cut-off values associated with increased mortality are unknown.

Methods and results

In patients with all-cause TR assessed both qualitatively and quantitatively by proximal iso-velocity surface area method, long-term and 1-year outcome analysis was conducted. Thresholds for excess mortality were assessed using spline curves, receiver-operating characteristic curves, and minimum P-value analysis. The study involved 676 patients with all-cause TR (age 73.9 ± 14 years, male 45%, ejection fraction 52.9 ± 14%). Effective regurgitant orifice (ERO) was strongly associated with decreased survival in unadjusted [hazard ratio (HR) 2.38 (1.79–3.01), P < 0.0001 per 0.1 cm2 increment] and adjusted [2.6 (1.25–5.0), P = 0.01] analyses. Quantitative grading was superior to qualitative grading in prediction of outcome (P < 0.01). The optimal cut-off value for the best separation in survival between groups of patients with severe vs. lesser degree of TR was 0.35 cm2 [P < 0.0001, HR =2.0 (1.5–2.7)]. ERO negatively impacted survival, even when including only the subgroup of patients with severe TR [HR 1.5 (1.01–2.3); P = 0.04]. The optimal threshold corresponding for the best separation for survival between groups of patients with severe vs. ‘torrential’ TR was 0.7 cm2 [P = 0.005, HR =2.6 (1.2–5.1)].

Conclusion 

TR can be severe and even ‘torrential’ and is associated with excess mortality. Quantitative assessment of TR by ERO measurement is a powerful independent predictor of outcome, superior to standard qualitative assessment. The optimal cut-off above which mortality is increased is 0.35 cm2, similar albeit slightly lower than suggested in recent guidelines. Torrential TR >0.7 cm2 is associated with poorer survival compared to patients with severe TR (ERO > 0.4 cm2 and <0.7 cm2).

Local recurrence after breast cancer affects specific survival differently according to patient age

PURPOSE

Young age is known to be an independent factor for developing local recurrence (LR) in breast cancer patients. It has also been shown that the occurrence of LR negatively affects patient outcome, especially if LR occurs within 3 years after treatment of the primary tumour. The question whether the impact of LR on patient outcome differs according to the patient's age has not been addressed before. The purpose of the present study is to investigate cancer-specific survival (CSS) as well as overall survival after LR in young patients (<50 years old) and to compare it to older patients. The age cut-off level was taken as 50 to avoid strong imbalance in patient numbers between the 2 groups.

PATIENTS AND METHODS

Between 1974 and 2003, 2,130 breast cancer patients were treated with conservative surgery and axillary dissection. All of them received post-operative radiotherapy. Adjuvant chemo- and/or hormonal therapy was given according to the prognostic factors and the treatment policy at the time of diagnosis. Only biopsy-confirmed ipsilateral LRs were taken into account. Early LRs were those observed within 36 months after surgery, and late LRs were those which occurred thereafter. The median follow-up was 100 months. Survival analysis was conducted with the Kaplan-Meier method.

RESULTS

The median age was 59 years. There were 472 patients aged <50 years versus 1,658 older patients. Pathological tumour size, hormone receptor status and lymph node involvement were evenly distributed in the 2 groups. The 5- and 10-year CSS was 92.3 and 83.9% in young patients, and 94.4 and 87.6% in older patients (p = 0.061), respectively. Overall, 200 LRs were observed; 52 of them (26%) were early LRs. The rate of LR was significantly higher in young patients: at 5 years, it was 10.5 versus 3.7% in patients ≥50 years; the respective rates at 10 years were 17.8 and 8.8% (p < 0.0001). The 5- and 10-year CSS in patients who developed LR was 86.8 and 76.0%, versus 94.7 and 88.2% in patients who did not develop LR (p < 0.0001). The 5-year CSS after LR in young and older patients was 77.6 and 65.7%, respectively (p = 0.028).

CONCLUSION

Although young patients experience more LR than older ones, once LR occurs, young patients have a better outcome than the others. Possible hypotheses are: (1) more aggressive treatment in young patients after LR; (2) the treatment is better sustained in young patients; (3) biological differences in the characteristics of LR.

Diversity in expression and prognostic significance of G1/S cyclins in human primary lung carcinomas

Expression of cyclin A, cyclin E and cdk2 was examined immunohistochemically in 144 cases of primary non-small cell lung carcinoma to evaluate their prognostic value. Cyclin A was co-expressed with cdk2 in the proliferating cells, ie those showing positive Ki-67 staining. The labelling index (LI) of cyclin A revealed a positive correlation with the S-phase fraction and an inverse correlation with histological differentiation. Furthermore, high cyclin A LIs indicated a poor prognosis in all histological types. Cyclin E exhibited a characteristic staining pattern: in squamous cell carcinoma (SCC), differentiated cells without Ki-67 staining revealed cyclin E positivity with expression of cdk2. Conversely, in adenocarcinoma (AC), proliferating cells revealed cyclin E positivity. Cases of large cell carcinoma showed heterogeneous cyclin E staining patterns, unlike those of SCC or AC. Statistical analyses also revealed a marked contrast between SCC and AC. In AC, the LI of cyclin E was inversely correlated with histological differentiation and a high LI predicted a worse prognosis. In contrast, in SCC, the LI of cyclin E correlated positively with histological differentiation and better prognosis. However, the expression levels of cyclin E mRNA evaluated by quantitative RT-PCR were higher in poorly differentiated SCC and AC, suggesting that protein turnover plays a large role in determining cyclin E protein levels. Although the expression of cyclins was demonstrated to be diversely regulated depending on the histological type, the combined immunohistochemical analyses performed in this study on these proteins could be useful tools for evaluating patient prognosis in lung carcinomas.

The Prognostic Value of Hormone Receptor Detection by Enzyme Immunoassay and Immunohistochemistry: A Prospective Study in Patients with Early Breast Cancer

The main reason to determine estrogen (ER) and progesterone receptors (PR) in breast cancer is their predictive value for the response to endocrine therapy. In addition, ER and PR are often used as prognostic indicators. Enzyme immunoassay (EIA) and immunocytochemical assay (ICA) are two methods for determining ER and PR. These two methods have not been compared with each other in relation to clinical endpoints. In the present study we prospectively evaluated the prognostic value of ER and PR as determined by ICA and EIA in 223 and 207 patients, respectively, with early breast cancer. ER was positive in approximately 77% of patients, while PR was positive in approximately 65% of patients. The proportion of potential agreement beyond chance between EIA and ICA was 0.58 and 0.65 for ER and PR, respectively. The median follow-up was 86 months. Both ER and PR appeared to be weak prognostic factors. There were no differences in prognostic value according to the time point of analysis or cutoff value chosen, nor were there any differences in the prognostic value of hormone receptors detected by ICA or EIA. Both methods appear to be equivalent in terms of qualification and prognostic value.

CDK-inhibitors-associated kinase activity: a possible determinant of malignant potential in smooth muscle tumors of the external soft tissue

There has been accumulating histological observation of leiomyoma and leiomyosarcoma of the external soft tissue regarding their differential diagnosis. The definitive diagnostic tools have not been established, however, nor have the pathological mechanisms of cell proliferation in these tumors been clarified. Herein, expression of the cyclin-dependent kinase inhibitors (CKIs), p21, p27 and p57 and their associated kinase activities were examined in 61 cases of soft tissue smooth muscle tumors. Immunohistochemical staining showed that all 3 inhibitor proteins were expressed in all cases of leiomyoma and leiomyosarcoma, but that the mean values of their labeling indices (LIs) were higher in the cases of leiomyosarcoma. In addition, the LIs of p21 and p27 were inversely correlated in total cases. Immunoblotting revealed that these proteins are expressed at higher levels in tumors, in particular, in leiomyosarcoma. When CKIs were immunoprecipitated from tissue extracts, cyclin/cdk protein complexes associated with, at least, 1 CKI were detectable only in tumor tissues. Furthermore, cdk2 or cdk4 kinase activity manifested by these cyclin/cdk/CKI complexes (CKI-associated kinase activity) was detectable exclusively from leiomyosarcoma, but not from leiomyoma. Among the cases of leiomyosarcoma, cdk2 activity was generally found associated either with p21 or p27, but not both. Statistical analysis indicated that p21- and p27 LIs are predictive of positive or negative clinical outcome, respectively. In conclusion, the participation of CKIs in active cyclin/cdk complexes in a reciprocal and redundant manner and subsequent CKI- associated kinase activity are the characteristic profiles of malignant phenotype in these tumors. Moreover, immunohistochemical detection of CKIs may provide a useful tool for evaluating patients' prognosis.

Problems in defining cutoff points of continuous prognostic factors: example of tumor thickness in primary cutaneous melanoma

Continuous prognostic factors are often categorized by defining optimized cutoff points. One component of criticism of this approach is the problem of multiple testing that leads to an overestimation of the true prognostic impact of the variable. The present study focuses on another crucial point by investigating the dependence of optimized cutoff points on the observed distribution of the continuous variable. The continuous variable investigated was the vertical tumor thickness according to Breslow, which is known to be the most important prognostic factor in primary melanoma. Based on the data of 5093 patients, stratified random samples were drawn out of six artificially created distributions of tumor thickness. For each of these samples, Cox models were calculated to explore optimized cutoff points for tumor thickness together with other prognostic variables. The optimized cutoff points for tumour thickness varied considerably with the underlying distribution. Even in samples from the same distribution, the range of cutoff points was amazingly broad and, for some of the distributions, covered the whole region of possible values. The results of the present study demonstrate that optimized cutoff points are extremely data dependent and vary notably even if prerequisites are constant. Therefore, if the classification of a continuous prognostic factor is necessary, it should not be based on the results of one single study, but on consensus discussions including the findings of several investigations.

Biological indicators of survival in patients treated by surgery for squamous cell carcinoma of the oral cavity and oropharynx

Squamous cell carcinomas of the head and neck are a heterogeneous group of tumours with regard to anatomical site, natural history and response to various treatments. Assessment of the role of biomarkers as indicators of prognosis or response to treatment is thus complex. In the last decade, different biomarkers have been investigated in the search for objective and reproducible indicators of prognosis. In 69 squamous cell carcinomas of the oral cavity or oropharynx from patients treated with radical surgery alone, we determined cell kinetics, evaluated as in vitro 3H-thymidine labelling index (TLI), p53, bcl-2 and glutathione S-transferase pi (GST pi) expression, by using immunohistochemical methods. The biological variables were unrelated to one another or to established clinical and pathological prognostic factors. Univariate analysis showed that a low proliferative activity was associated to a significantly higher risk of death than that observed in patients with a high TLI, whereas p53, bcl-2 and GST pi expression did not provide prognostic information. Multivariate analysis showed that cell proliferation, gender and nodal status retained their clinical relevance. In the subset of node-negative patients, TLI and p53 expression were indicators of survival. Moreover, the combined analysis of TLI and p53 expression identified a subgroup of node-negative patients with slowly proliferating and highly p53-expressing tumours who died within 1 year of radical surgery. These results indicate that in patients with operable oral cavity and oropharyngeal cancer, biomarkers can provide important information on clinical outcome.

Prognostic significance of S-phase fraction detected by antithymidine antibodies in epidermoid cervix carcinomas

PURPOSE

To assess the predictive value of pretreatment proliferative activity of epidermoid cervix carcinoma cells with respect to short- and long-term results of radiotherapy.

METHODS AND MATERIALS

The proliferative activity of 25 epidermoid cervix carcinomas was evaluated as the immunofluorescent labeling index (LI) by rabbit antithymidine antibodies reacting specifically with single-stranded DNA of replication forks in S-phase cells. The short-term clinical outcome was estimated at 3-6 months after treatment by visual and palpatory examination. Three-year follow-up data were obtained through hospital charts and correspondence with referring physicians for only 19 patients.

RESULTS

There was no statistically significant association between LI and such conventional prognostic factors as clinical stage. The LI value of cervix carcinomas was significantly associated with complete regression at 3-6 months after radiotherapy and 3-year disease-free survival. Complete regression at 3-6 months was observed in 87.5% patients with fast proliferating tumors (LI > 7.0%), and only in 41.2% patients with slowly proliferating tumors (p = 0.03). Probability of 3-year disease-free survival was 85.7% in patients with fast proliferating tumors and 50.0% in those with slowly proliferating tumors (p = 0.05).

CONCLUSIONS

The immunofluorescent LI of epidermoid cervix carcinoma is able to provide prognostic information on short-term tumor response to radiotherapy and disease-free survival.

The relationship of estrogen receptor status to DNA ploidy in breast cancer

The relationship of estrogen receptor (ER) status to DNA ploidy was investigated in 121 patients with breast cancer who underwent surgery. Lymph node status was evaluated histologically and ER levels were determined by the dextran-coated charcoal method, with a level of 3 fmol/mg.protein being considered positive. Flow cytometric DNA content was analyzed using paraffin-embedded tissue blocks. Sixty-three percent of the specimens were ER+, while 37 percent were negative. Sixty-one patients (50.4 percent) were diploid and 60 aneuploid. A statistically significant correlation between the ER positivity rate and diploid DNA was found. Higher ER levels were seen in the postmenopausal patients with diploid tumors than in those with aneuploid tumors and there was a significant tendency for ER levels to be higher in the diploid tumors. Nodal status was not correlated with ER positivity or ploidy pattern. The present results indicate that ER levels are correlated with DNA ploidy, and reflect the degree of functional differentiation.

Flow cytometry in primary breast cancer: improving the prognostic value of the fraction of cells in the S-phase by optimal categorisation of cut-off levels

The use of continuous prognostic variables is clinically impractical, and arbitrarily chosen cut-off points can result in a loss of prognostic information. Here we report findings from a study of primary breast cancer, showing how the prognostic value of the fraction of cells in the S-phase of the cell cycle (SPF), as measured by flow cytometry, can be affected by the SPF cut-off level(s) adopted. It was possible to evaluate the SPF in 566 (94%) of 603 consecutive cases where fresh frozen specimens were available in a tumour bank at our department. Clinically, all patients were without distant spread at the time of diagnosis, and the median duration of follow-up was 4 years. Using different survival end-points and chi 2 values for each cut-off level, two optimal cut-off points, at the 7% and 12% levels, were consistently obtained for the SPF. Furthermore, both disease-free survival and the relative risk of recurrence exhibited a non-linear relationship with SPF values; the curves implied that the prognosis was better among patients with SPF values about 2-5% than in patients with lower SPF values (parabolic shape), though the relationship with higher SPF values approached linearity. The non-linearity of the curves is incompatible with the general use of the median SPF as a prognostic cut-off value. An alternative procedure might be to use two cut-off levels, one to distinguish patients with the lowest SPF values (i.e. within the parabolic survival curve) from those with higher values (i.e. with a survival curve approaching linearity), the other to distinguish between patients with intermediate SPF values and those with high values (i.e. within the almost linear part of the survival curve). The 7% and 12% obtained here would be suitable for this purpose. We conclude that prognostic information can be gained by dividing the SPF into three prognostic categories (less than 7.0%, 7.0-11.9% and greater than or equal to 12%), instead of using the median SPF level.

Indicators of prognosis in node-negative breast cancer

Measures of the proliferative activity of tumor cells have prognostic value in patients with node-negative breast cancer. We studied 367 women in southern Sweden who had undergone surgical resection for such cancer. Tumor specimens were analyzed with DNA flow cytometry in order to estimate both the DNA content (ploidy) and the fraction of cells in the synthetic phase of the cell cycle (S phase). The median duration of follow-up was four years; 28 percent of the patients received adjuvant therapy, usually with tamoxifen (n = 83). A multivariate analysis based on complete data on 250 patients included the following covariates: age (greater than or equal to 75, 50 to 74, and less than or equal to 49 years), tumor size (less than or equal to 20 vs. greater than 20 mm), concentration of estrogen and progesterone receptors (less than 10 vs. greater than or equal to 10 fmol per milligram of protein), ploidy (diploid vs. nondiploid), and S-phase category (fraction of cells in S phase: less than 7.0 percent, 7.0 to 11.9 percent, and greater than or equal to 12 percent). The S-phase fraction yielded the most prognostic information, followed by progesterone-receptor status and tumor size. A prognostic model based on these three variables identified 37 percent of the patients as constituting a high-risk group with a fourfold increased risk of distant recurrence. In the remaining 63 percent of the patients, the five-year overall survival rate (92 +/- 4 [+/- SE] percent) did not differ from the expected age-adjusted rate for Swedish women. We conclude that a prognostic index that includes indicators of the proliferative activity of tumor cells may be able to identify women with node-negative breast cancer in whom the risk of recurrence is sufficiently low that adjuvant chemotherapy can be avoided.

Expression of glutathione S-transferase B1, B2, Mu and Pi in breast cancers and their relationship to oestrogen receptor status

The concentrations of glutathione S-transferase (GST) B1 and B2 (Alpha), Pi and Mu have been measured by radioimmunoassay in cytosols from 28 oestrogen receptor (ER) rich an 30 ER-poor breast tumours. GST B1, B2 and Pi was detected in all 58 breast tumour cytosols whilst GST Mu was found in only 28. Of the GSTs, Pi was expressed most strongly in all cytosols and the concentration was significantly higher in ER-poor tumour cytosols than in ER-rich tumours (P less than 0.01). As with GST Pi, the highest levels of GST B1 and GST B2 were found in ER-poor tumour cytosols; the levels of GST B1 and GST B2 were positively correlated (r = 0.66, P less than 0.001). No quantitative or qualitative association was found between ER status and GST Mu which was expressed in 46% of ER-rich and 50% of ER-poor tumour cytosols. No relationship could be found between GST expression and age, menopausal status, lymph node involvement or tumour T stage in the subgroup of patients in whom this information was available. These data suggest that a common mechanism is responsible for GST induction in ER-poor tumours and that the nulled Mu phenotype has no increased susceptibility to developing breast cancer.

Circulating tumor marker levels in advanced breast carcinoma correlate with the extent of metastatic disease

The importance of the extent of metastatic disease in the circulating levels of CA 15-3 and carcinoembryonic antigens (CEA) was studied in 173 patients with advanced breast carcinoma. Estimates of the extent of metastatic disease were obtained by an objective arbitrary scale. Patients were observed clinically after serum samples were obtained, and survival was recorded. Elevated values of CA 15-3 (greater than 40 U/ml) were seen in 130 patients and CEA values (greater than 5 ng/ml) in 97 cases (75% versus 56%, P less than .0001). Elevated CA 15-3 values correlated with the estimated extent of metastatic disease (P less than .0001), number of metastases (P = .0006), and survival from study entry (P = .01). Elevated CEA values correlated with extent of disease (P less than .0001), but not with the number of metastases or with survival. No correlation was found between the elevated values of CA 15-3 or CEA and age, menopausal status, and initial tumor size or nodal status. The combination of the elevated values of CA 15-3 and CEA was more sensitive than CA 15-3 alone (P = .04), but there were no significant improvements when subgroups were considered. Significant differences, that depended on which specific organ was affected dominantly by metastases, were seen in the mean levels of CA 15-3 antigen. Similarly, patients with liver involvement had higher mean levels of CA 15-3 than those without hepatic metastases. A stepwise regression analysis of the dominant site of metastases, liver involvement, and estimated extent of disease showed that only the latter parameter retained a significant correlation with CA 15-3 antigen levels (P less than .0001). Median survival of patients who showed abnormal CA 15-3 levels was significantly shorter than that of patients with nonelevated CA 15-3 (10.1 versus 18.0 months, P = .04). This difference was not appreciated with CEA levels (10.2 versus 12.2 months, P = .4). We conclude that tumor marker levels in patients with advanced breast carcinoma correlate with the extent of metastatic disease. In addition, the CA 15-3 assay is more sensitive and correlates more accurately with the extent of disease than CEA. Finally, the observed CA 15-3 differences by organ involvement are related to the extent of disease variations. The objective evaluation of the extent of metastatic disease provides a new approach in the study and comparison of breast cancer-associated tumor markers.

Cell proliferation kinetics in human solid tumors: relation to probability of metastatic dissemination and long-term survival

A large number of studies have investigated the relationship between the long-term survival and the percentage of tumor cells in S phase assessed by autoradiography after tritiated thymidine labelling, image cytometry, flow cytometry or labelling with an halogenated analog of thymidine, in various types of human solid tumors. The survey of the results clearly shows that the S-phase fraction (SPF) is of high prognostic significance in several types of cancers, in particular in breast cancers, non-Hodgkin lymphomas, ovarian cancers, neuroblastoma, bladder cancers and lung cancers. SPF was found of high independent significance in 10 of the 11 studies in which multivariate analyses of prognostic factors had been carried out. Proliferation appears generally to be of higher prognostic significance than ploidy. In view of the wide differences in the biological characteristics of the tumors studied, it is likely that the association between a high proliferation rate and the degree of tumor aggressiveness is a general feature of human solid tumors. However, high proliferative rate of tumor cells is probably not the cause of tumor biological aggressiveness but a variable associated with it. The extent to which cells escape from the regulatory systems which control their proliferation appears to be a good index of tumor progression.

The labelling index: a prognostic factor in head and neck carcinoma

The thymidine labelling index (LI), representing the percentage of cells in the DNA-synthesis phase, was measured in vitro prior to therapy in 87 patients with squamous cell carcinoma of the head and neck, who were treated between 1977 and 1982. The LI was not related to patient age, site of the tumour, clinical stage or histological grade. Overall survival was 44.5%. Univariate analysis demonstrated that survival was affected by the following factors: (1) age: patients older than 55 had a better outcome (p = 0.03); (2) site of the tumour (p = 0.005): laryngeal tumours had the best survival; (3) clinical stage (p = 0.05). Histological grade did not influence the survival (p = 0.41). Patients having a tumour LI higher than 15.5% (mean + 1 S.D.) had a significantly lower survival than patients with lower tumour LI (p = 0.008). A multivariate analysis using the Cox model showed that clinical stage and LI kept their prognostic impact with regard to survival. Finally, survival after relapse was lower in patients with a high tumour LI. These results demonstrate that a high tumour proliferation rate is an additional factor influencing the disease outcome in head and neck carcinoma. Patients with bad prognosis defined by this parameter could be offered a more energetic treatment.