Purpose: Although serum tumor markers (STMs) are widely used in clinical practice, their predictive role for the response to anticancer treatment is still controversial. The correlation of CEA, CA 15.3, CA 19.9, CA 125 (only with peritoneal involvement) and NSE levels with imaging response and clinical benefit was investigated in 60 non-selected patients with metastatic epithelial cancers treated by single-agent docetaxel chemotherapy.
Methods: STM measurement was performed at baseline and subsequently every three to four weeks. We applied the WHO criteria to evaluate both STM and instrumental responses. Concordance analysis was performed by the Cohen Kw index, and the significance of the results was established using the Fleiss, Cohen & Everitt test. Qualitative interpretation of data was obtained with the Landis & Koch scale. Correlations of STM response with clinical benefit (PS or pain improvement) were evaluated by the chi-square test.
Results: The primary tumors included breast cancers (38 patients), gastrointestinal non-colorectal cancers (12 patients), and lung cancers (10 patients). An overall significant good degree of agreement was observed between STM and instrumental response (p < 0.0005). The degree of agreement for each marker was as follows: excellent for CEA (p < 0.0005) and CA 125 (p = 0.006), good for CA 15.3 (p < 0.0005) and CA 19.9 (p = 0.011). Restricted analysis for the correlation of each marker with primary tumor origin showed good prediction of radiological response for CA 15.3 and CEA in breast cancer patients (p<0.0005 for both), for CEA and CA 19.9 in gastrointestinal cancer patients (p = 0.01 and 0.04, respectively), and for CEA+NSE in lung cancer patients (p = 0.01). Conversely, STM response did not correlate significantly with the clinical benefit for the patients, both in terms of PS and pain improvement (p = 0.24 and p=0.42, respectively).
Conclusion: This study showed STMs to be good predictors of tumor response. Although STMs cannot replace diagnostic imaging, in metastatic cancer they might be useful to optimize the timing of radiological re-evaluation in the palliative setting.