Renal cell carcinoma (RCC) accounts for 2–3% of all adult cancers and represents the third most common urological malignancy in Europe (1). At diagnosis, one-third of patients present with locally advanced or metastatic disease, and one-third of patients undergoing nephrectomy will eventually develop metastasis (2,3). Previously, immunotherapy agents, such as interleukin-2 and interferon (IFN)-α, were the only treatments available and achieved response rates of ~10–22% (4–9). In recent years, however, the strategy for treating metastatic RCC (mRCC) has changed from immunotherapy to the administration of molecular-targeted therapies, such as multitargeted inhibitors of tyrosine kinases, and mammalian target of rapamycin. Therefore, the establishment of a tool for predicting the effect of targeted agents for mRCC is of critical importance.

One of the most well-established classification systems for patients with mRCC is the Memorial Sloan Kettering Cancer Center (MSKCC) system reported by Motzer et al in 1999 (10) and modified in 2002 (11). This model was independently validated by investigators at the Cleveland Clinic and has been used for the study and interpretation of cytokine and targeted drug therapies (12). In this era of molecular-targeted therapy, prognostic factors for mRCC other than the MSKCC risk classification system have been identified, such as the serum C-reactive protein (CRP) level (13,14), metastatic status (15,16) and tumor shrinkage (17). However, there have been few reports on independent prognostic factors.

To further investigate the association between clinical parameters and overall survival (OS) in mRCC, a retrospective analysis of consecutive patients treated with molecular-targeted therapy at the Kyushu Cancer Center (Fukuoka, Japan) was performed.

Molecular-targeted therapy has markedly changed the treatment strategy for mRCC, and several recent studies have investigated the clinical prognostic factors. At present, the most widely used system is the MSKCC classification, which may facilitate prognostic individualization in mRCC patients who received systemic therapy (11). However, despite being validated in the era of molecular-targeted therapy (20), this model was developed based on data from patients treated with IFN-α in a clinical trial. Therefore, there is a need for a system that allows for a more precise assessment of the prognostic risk in patients with mRCC receiving molecular-targeted therapy, since such a tool would be useful for counseling patients, evaluating therapeutic options and planning treatment. However, only few reports have identified any independent prognostic factors. Therefore, the aim of the present study was to retrospectively investigate the prognostic factors for patients with mRCC treated with molecular-targeted agents.

The current estimates for the median OS in patients with mRCC range from 22.9 to 26.4 months (21,22). The OS of 23.7 months in the present study falls within this range (Fig. 1), indicating that molecular-targeted agents have been administered appropriately at our institution. The median OS stratified by the MSKCC risk classification was 28.5, 20.9 and 8.1 months for the favorable-, intermediate- and poor-risk groups, respectively (P=0.137; degree of freedom: 2). While the OS of the favorable-risk group was longer compared with that of the intermediate- and poor-risk groups, no significant difference in the OS was found between the favorable- and intermediate-risk (P=0.2713), the favorable- and poor-risk (P=0.0664), or the intermediate- and poor-risk groups (P=0.1426). Therefore, univariate and multivariate analyses were retrospectively performed using the Cox proportional hazards model in consecutive patients treated with molecular-targeted therapy, to determine the association between the OS and clinical parameters, including five risk factors of the MSKCC risk classification (Table III). Univariate analyses for various factors identified prior nephrectomy, number of metastatic sites, anemia, best response to first-line treatment and PFS with first-line treatment as prognostic variables. Furthermore, multivariate analyses identified the following as independent prognostic factors: Number of metastatic sites (2: HR=3.351, 95% CI: 1.460–8.201, P=0.004; ≥3: HR=6.397, 95% CI: 1.939–20.209, P=0.003) and time from diagnosis to therapy (≥1 year: HR=0.334, 95% CI: 0.137–0.755, P=0.008) as pre-treatment factors, and PFS with first-line treatment (≥5.1 months: HR=0.353, 95% CI: 0.156–0.766, P=0.008) and number of lines of molecular-targeted agents (≥3: HR=0.248, 95% CI: 0.091–0.664, P=0.006) as post-treatment factors.

The pre-treatment factor ‘time from diagnosis to therapy’ is included in the MSKCC system, and subsequent studies conducted to validate the prognostic factors for RCC have evaluated the time from diagnosis to the initiation of systemic therapy (23–25). However, a survival analysis stratified by the number of disease sites is not often performed in clinical trials, although it may represent an additional prognostic factor of outcome. Grassi et al reported that the presence of >2 disease sites was associated with a statistically significantly shortened PFS and OS (26). The number of metastatic sites may thus be a surrogate for the tumor burden, which may be easily evaluated, although it does not include the spread of metastases.

Currently, the most widely used prognostic factor model is based on the MSKCC (11). Adverse prognostic factors in a multivariable analysis included a low Karnofsky performance status (<80%), high LDH (>1.5 times the upper limit of normal), low serum hemoglobin levels, high corrected serum calcium levels (>10 mg/dl), and a time from initial diagnosis to treatment of <1 year. Based on these five risk factors, each patient was assigned to one of three risk groups: Favorable risk (0 risk factors), intermediate risk (1–2 risk factors) and poor risk (≥3 risk factors). This means that the MSKCC risk is classified based only on the number of factors present, not each risk factor or any combination. Therefore, the breadth of cases included, particularly in the intermediate- and poor-risk groups, is wide, and variations naturally exist among patients, even within each risk group. As such, urological oncologists recognize that even patients in the same risk group may not achieve the same results. From this standpoint, it may be argued that post-treatment as well as pre-treatment variables are important as prognostic factors.

In the present study, a significant difference was observed, not in the best response to first-line treatment, but in PFS with first-line treatment (≥5.1 months: HR=0.353, 95% CI: 0.156–0.766, P=0.008) in the multivariate analysis, and these results were naturally obtained after molecular-targeted agent administration. Recent analyses have demonstrated that patients with insufficient response to first-line treatment have a dismal prognosis (27,28). Siedal et al suggested that early tumor shrinkage is a prognostic tool, and superior tumor shrinkage is associated with a favorable prognosis (17). The patients in that analysis were stratified into five groups according to the change in the tumor size at the first-treatment evaluation, whereas the present study stratified patients into three groups according to the best response to first-line treatment. However, the best response to first-line treatment was not found to be an independent prognostic factor in the present study. This result suggested that the clinical response rate may not reflect the survival time. In another study, Seidal et al suggested that a PFS with first-line treatment of >6 months was an independent prognostic marker (29). Heng et al also mentioned that a PFS of 6 months under first-line vascular endothelial growth factor-targeted treatment was applied as a cut-off marker and proved capable of significantly differentiating patients with favorable and poor prognosis (30). These descriptions are consistent with our observations in the present study, although the duration of PFS with the first-line treatment was slightly different in the previous study. The longer PFS with first-line treatment may thus be associated with the better prognosis observed in patients with mRCC. These results appear to be important, not only for urological oncologists, but also for patients in clinical practice; thus, even patients in the same risk group may achieve a different OS compared with what was expected prior to treatment administration. Given that urological oncologists naturally interact with patients from pre-treatment to post-treatment, they have several opportunities to explain to their patients their condition and prognosis. Urological oncologists may therefore be able to discuss a patient's prognosis with greater specificity after treatment administration. This information may greatly help patients make important decisions.

In addition, multivariate analyses also identified the number of lines of molecular-targeted agents (≥3: HR=0.243, 95% CI: 0.089–0.654, P=0.005) to be an independent prognostic factor. Ko et al demonstrated that patients with mRCC who were able to receive more lines of molecular-targeted therapy lived longer, with longer PFS (31). These results suggest that sequential therapy with molecular-targeted agents may prolong the survival of patients with mRCC.

Multiple candidate predictive and prognostic biomarkers have been evaluated (32–38). However, the association between the OS and these biomarkers was not examined in these previous studies. At present, no available biomarkers are superior to clinical parameters, such as those used for the MSKCC score.

The limitations of such an analysis are its retrospective nature and the small number of patients enrolled. However, our experience from everyday clinical practice has highlighted the potential use of such information on the prognostic role of PFS with first-line treatment with molecular-targeted therapy for mRCC. The results of the present study indicate that the PFS of first-line treatment may be a meaningful intermediate endpoint for OS in mRCC patients treated with molecular-targeted therapy.