Cyclooxygenases (both isoforms, COX-I and COX-II) oxidize arachidonic acid to prostaglandin H2, which is converted by different synthases to prostaglandin-E2, -D2, -I2, -F2α, and thromboxane A2. These prostaglandins inhibit apoptosis and promote cell division, metastasis and angiogenesis leading to increased tumor growth [1]. An antiangiogenic/antiinflammatory therapy with COX-II inhibitors and pioglitazone combined with metronomic low-dose chemotherapy with either capecitabine or trofosfamide seems to be well tolerated and promising in patients with advanced carcinomas [2, 3]. However, in one study in patients with colorectal cancer increased gastrointestinal toxicity was reported [4]. Since COX-II inhibitors are known to elicit renal side effects to a similar extent than conventional non-steroidal antiinflammatory drugs [5], the detailed analysis of any change in serum creatinine concentrations and glomerular filtration rate in a prospective trial of antiangiogenic/antiinflammatory therapy in advanced cancer was our primary objective.

We found a significant increase in mean serum creatinine concentration (p < 0.0001) already in the first month after initiation of treatment with rofecoxib. This increase was sustained during treatment and ranged around 0.15 mg/dl, corresponding to a decrease of eGFR of around 10 ml/min, reaching up to 14 ml/min in month 4. However, mean serum creatinine concentration remained within the normal range during and after end of the trial and no dialysis treatment was needed. Rofecoxib is known to cause acute renal failure at a rate of 1.2 to 1.5% in trials studying several thousands of patients [7, 8], but even a single dose of rofecoxib can cause severe acute renal failure [9]. Furthermore, the half life of COX-II inhibitors rofecoxib and celecoxib is prolonged in liver failure [10]. Though patients with severe hepatic failure were excluded from the present study, 52% of the patients presented with liver metastasis and thus might have been subjected to a higher exposure of rofecoxib due to prolonged half life. Progressive underlying malignant disease may as well have contributed to functional prerenal failure i.e. due to decreased fluid intake. The decreased renal function however tended to normalize within 2 months after cessation of rofecoxib therapy (Figure 2 and 4), suggesting rofecoxib as the more likely cause of decreased renal function, especially as we observed no severe nausea, vomiting or diarrhea during the treatment phase.

We observed edema formation in 30 patients (34%), which is above the rate observed in large studies ranging between 7.7 and 23.3% [8, 11, 12] and which is possibly due to the persistent decrease in renal function, the well-known sodium- and/or fluid retaining properties of COX-2 inhibitors and thiazolidindiones, and the underlying malignant disease with hypoproteinemia.

The rise in mean systolic blood pressure during therapy corresponds to published data [13, 14] and went along with increasing body weight, probably both reflecting renal impairment and subsequent fluid retention. Weight gain is often seen in patients with diabetes treated with pioglitazone [15] and pioglitazone may have contributed to edema formation and the observed increase in body weight in our study, but weight gain in our study population is even more astonishing in regard of the underlying malignancy which usually leads to progressive weight loss as demonstrated after end of treatment.

The antiinflammatory antiangiogenic therapy with a COX-II inhibitor in combination with pioglitazone and either trofosfamide or capecitabine seems to be both well tolerated and promising with primary disease response rates reaching up to 19% [3].

This study was stopped because of withdrawal of rofecoxib from the market, but the concept of an antiinflammatory antiangiogenic therapy with COX-II inhibitors proves to be true also for other COX-II inhibitors like celecoxib, still available on the market [16–18]. There are also clearly class effects like prolonged half life in liver failure [10] or decreasing renal function, which have been shown for rofecoxib as well as for celecoxib [19].

The decrease in renal function in our study has to be seen in the clinical context. A rise in mean serum creatinine concentration of 0.15 mg/dl is of limited importance to young patients with normal body weight and normal renal function. However, a mean fall in eGFR of up to 14 ml/min nevertheless may urge the physician to adapt the dose of concomitant medications or even stop the treatment with a COX-II inhibitor especially in the elderly patient with reduced body weight and pre-existing renal failure, suggesting the report of our results to be of persistent importance even in the light of withdrawal of rofecoxib from the market.

We recommend close monitoring of renal function and blood pressure of patients with cancer treated with COX-II inhibitors, especially in a setting with pre-existing renal and hepatic failure.