Colorectal cancer treatment and early response evaluation : how do we best evaluate treatment response?

Abstract

Colorectal cancer (CRC) is common, being responsible for around 12% of new cancers in Sweden, and contributing heavily to the large numbers of cancer deaths yearly. CRC is the second most common cause of cancer death not only in Sweden but also worldwide.

In a large (n=567) multicentre, phase III study, the Nordic VI study, two different ways of giving 5-FU, bolus (FLv) and protracted infusion (Lv5FU2), together with irinotecan, in patients with metastatic (m)CRC was tested without any differences in progression-free survival (PFS) [9 months] or overall survival (OS) [19 months]. Fewer objective responses were seen in the FLIRI group (35% versus 49%, p = 0.001), but the metastatic resection rate did not differ (4% versus 6%, p = 0.3).

In the same study a subset (n=51) of the population was evaluated for early metabolic treatment response with [18F]-2-fluoro-2-deoxy-D-glucose (FDG) positron emission tomography (PET). The mean baseline standard uptake value (SUV) for all tumor lesions per patient was higher in non-responders than in responders (mean 7.4 versus 5.6, p = 0.02). There was a strong correlation between metabolic response (changes in SUV) and objective response (p = 0.00001), with a sensitivity of 77% and a specificity of 76%. There was no significant correlation between metabolic response and PFS (P = 0.5) or OS (p = 0.1).

In the Nordic VI study, selected tumor markers at baseline and during treatment were evaluated for their ability to predict response rate (RR), PFS and OS in two sub-studies. In the first one (n=90), low levels of tissue inhibitor of metallo-protease 1 (TIMP-1) levels at baseline were correlated to a higher probability of obtaining an objective response (P =0.007). Plasma TIMP-1scored as a continuous variable on a log scale (loge) was significantly associated with OS (P < 0.0001) and PFS (p =0.048).In the second sub-study (n=106) a significant correlation to OS was seen for baseline levels of all selected markers. In multivariate analyses with clinical parameters, TPA, CRP, SAA and TIMP-1 provided independent information. Changes during treatment, recorded as the slope gave with the exception of CA19-9 for OS less information about outcomes. The best correlation to response was seen for CEA, CA19-9 and TPA with AUC values of 0.78, 0.83 and 0.79, respectively, using a combined model based upon an interaction between the slope and the baseline value.

Health related quality of life (HRQoL) was also evaluated in a subset of the Nordic VI population (n=220). There were no differences in HRQoL between the two treatment groups at any time point. Emotional functioning and pain improved, and diarrhea worsened with time. Most baseline QoL subscales correlated with OS. Independent information on OS, but not PFS or RR was seen for physical functioning (p=.000), appetite (p=.028) and constipation (p=.041) together with hemoglobin level. A summary score, based on the sum of all scale items, was independently related to OS (p=.000) and PFS (p=.006) but not to RR. Most patients with an objective tumor response or a long (≥ 4 months) disease stabilisation had a favourable HRQoL outcome; however, a minor portion did not. No significant correlations were seen between changes in QoL parameters during treatment and RR, PFS and OS.