The alkylating agent cyclophosphamide (CYC) is a cytotoxic drug widely given, by either intravenous (IV) or oral routes of administration, to treat solid and hematologic malignancies (1) and rheumatic diseases, e.g., systemic lupus erythematosus, systemic sclerosis, or systemic necrotizing vasculitis (SNV) (2). It has long been recognized that CYC has substantial adverse effects, such as infections, secondary hematologic cancers, and urotoxicity. CYC-related urotoxicity includes hemorrhagic cystitis (3–5) and urinary tract cancers of the bladder (6, 7) and, less commonly, the ureter and renal pelvis (8–10) that may occur long after the first CYC exposure (2). When CYC was used to treat non-Hodgkin's lymphoma, cumulative 5-year hemorrhagic cystitis incidence rates of 12% and 12-year bladder cancer incidence rates of 11% (11) were reported, and the close link between cumulative CYC dose and secondary bladder cancer risk was established (12).
In the setting of SNV, CYC-related urotoxicity has also raised concerns (13, 14). In Wegener's granulomatosis (WG), an SNV frequently characterized by a chronic relapsing course requiring repeat treatment and for which CYC is a mainstay of therapy (15, 16), elevated urinary tract cancer and/or hemorrhagic cystitis rates have been demonstrated in previous studies (13, 14, 17–22). However, those studies showed wide variations in the frequency of urinary tract cancer and/or hemorrhagic cystitis, left unclear whether there is a cutoff dose below which cumulative CYC has no urotoxic effect, and did not provide information on CYC-related urotoxicity for SNV other than WG. While the results of previous studies tended to support a more favorable safety profile of intermittent IV CYC, as opposed to the originally described daily oral regimen (15, 16, 23), whether the administration route also affects CYC urothelial toxicity has not been evaluated (2).
In this retrospective study, we assessed the incidence of, and risk factors for, urinary tract cancer and/or hemorrhagic cystitis in CYC-treated patients with WG or a related small-vessel or medium-sized–vessel SNV, namely, microscopic polyangiitis (MPA), Churg-Strauss syndrome (CSS), and polyarteritis nodosa (PAN).
- Top of page
- PATIENTS AND METHODS
- AUTHOR CONTRIBUTIONS
Among the 805 CYC-treated SNV patients followed up for an average of 5.3 years, 7 developed urinary tract cancer (i.e., 6 had bladder cancer and 1 had ureter cancer), and 22 experienced hemorrhagic cystitis. The observed number of urinary tract cancer cases indicated a 5-fold higher incidence compared to the general population. Our analyses of risk factors, which were based on a nested sample, suggested that cumulative CYC dose, ever-oral CYC administration, and WG diagnosis were independent risk factors for urinary tract cancer and/or hemorrhagic cystitis. According to our univariate analyses, current or past tobacco smoking and a previous hemorrhagic cystitis episode also predicted urinary tract cancer.
Acrolein, a CYC metabolite excreted in the urine, is known to be responsible for CYC urotoxicity. Studies in animals have demonstrated that acrolein initiates bladder cancer (43) and hemorrhagic cystitis in a concentration- and time-dependent manner (44, 45). Further support for the notion of acrolein urotoxicity came from observations indicating a protective effect of adjunctive mesna, which neutralizes acrolein, against hemorrhagic cystitis in patients treated with high-dose IV CYC for hematologic malignancies (46–48) and against bladder cancer in rodents (49). Notably, our findings and a previously calculated summary estimate based on studies of WG, rheumatoid arthritis, and lymphoma patients (2) showed that prior hemorrhagic cystitis significantly multiplied the risk of subsequent urinary tract cancer by a factor of 5–7. This clustering of hemorrhagic cystitis with urinary tract cancer requires further investigation but may support the notion that these 2 events are pathogenetically linked.
The 2.4% estimated 10-year cumulative incidence and 5.0 SIR for urinary tract cancer are consistent with previously published bladder cancer incidence rates in patients with WG (21, 22) and a mixed population of patients with WG or MPA (19). As emphasized previously (19, 22), these data contrast with the higher 5% 10-year incidence rate and 30–31 SIR for bladder cancer observed in another WG cohort (17, 18). It is likely that these discrepancies can be explained, in large part, by the higher cumulative CYC dose (median 75 gm) received in the latter cohort. For our nested subgroup with available detailed CYC exposure data, the mean CYC dose was 25 gm. In addition, our findings confirmed the wide range of time intervals from first CYC exposure to urinary tract cancer development and further underscored the recommendation for prolonged and perhaps lifelong surveillance of CYC-treated SNV patients with periodic urinary cytology screening (50).
We could not reproduce the previously proposed values of 25 gm (21), 33 gm (22), or 100 gm (18) as thresholds associated with increased bladder cancer risk in WG. The wide overlapping of total cumulative CYC doses given to patients who experienced urotoxic events versus their counterparts (Figure 2) rather supports the hypothesis that urotoxicity cannot be predicted based on this parameter alone. That 2 cases of urinary tract cancer occurred in patients with seemingly modest cumulative CYC doses of 8 gm and 12 gm (Figure 2) highlights the likely influence of other risk factors. Whether tobacco smoking, which is a major contributor to bladder cancer (51), acts synergistically with CYC to induce bladder cancer is unknown but should be considered in the urotoxic risk assessment of CYC therapy. While a urotoxic threshold may not exist, our data are consistent with a sublinear dose-response profile with a less steep risk increase at lower, as opposed to higher, cumulative CYC exposures.
Perhaps our most original observation is that the route of CYC administration affected the risk of a urotoxic event. Even when adjusted for cumulative CYC dose, our results indicated that ever-oral CYC recipients had a shorter time to urotoxic events than patients treated exclusively with IV CYC. This observation, which could be made due to the wide use of IV CYC for SNV patients, has biologic plausibility in that the greater urotoxicity of daily oral CYC exposure might reflect prolonged urothelium contact with acrolein. However, this finding could also be partly explained by the more common practice of administering mesna and/or hyperhydration in combination with IV CYC. Oncology study results have shown that hyperhydration has protective efficacy against hemorrhagic cystitis comparable to that of mesna (46–48), but those evaluations were carried out in the setting of IV CYC and leave unresolved whether their results can be extrapolated to oral CYC.
Our study also provides new insight into the question of CYC-related urotoxicity in SNVs other than WG. Our SIR estimates indicated a slightly higher incidence of urinary tract cancer in WG patients than in those with non-WG diagnoses (i.e., MPA, CSS, or PAN). Moreover, our multivariate analyses retained WG diagnosis as an independent predictor of urinary tract cancer and/or hemorrhagic cystitis. In light of the higher cumulative CYC doses and its more common oral administration in WG patients, this finding may have resulted from residual confounding, with the variable WG carrying supplemental information on the burden of high, orally administered CYC exposure. However, we cannot exclude the possibility that the risk of urotoxic adverse events is higher in WG due to other, CYC-independent mechanisms.
Limitations of our study include that the identification of hemorrhagic cystitis and urinary tract cancer cases relied on a spontaneous reporting system. Therefore, our study might have underrated their true incidences. In contrast, the SIR for urinary tract cancer could also represent an overestimation because this calculation used general population incidence data on bladder cancer that did not include noninfiltrative tumors and because 1 of the diagnosed urinary tract cancer cases concerned the ureter. However, our inclusion of noninfiltrative bladder cancers in the numerator should have influenced only the time for these tumors to become infiltrative, whereas, due to their uncommon occurrence when compared to that of bladder cancer (52), the numbers of expected cancers of the ureter or the renal pelvis in our study population were only minimal. Interpretation of our SIR estimates must also consider the fact that they do not account for potential deviations in exposures to smoking or other etiologic urinary tract cancer risk factors between the SNV patients and the general population. Use of SNV patients who were not exposed to CYC as the reference group could have partly circumvented the latter shortcoming, but accurate urinary tract cancer incidence rates are unrealistic to ascertain for this population.
Moreover, our discussed findings on risk factors, i.e., CYC dose and administration route, reached statistical significance for the combined hemorrhagic cystitis and/or urinary tract cancer end point but not for urinary tract cancer alone. Thus, in light of the accumulating evidence that hemorrhagic cystitis might predict urinary tract cancer and considering that the point estimates of these variables consistently pointed in the same direction for either outcome, we think that the combined urotoxic end point has both clinical and methodologic relevance. Due to the small number of urotoxic events observed in this cohort and in previous studies (2), our study also highlights the general complexity of exploring this topic in this particular setting of vasculitis because of limited statistical power and precision, and the constrained possibilities to perform multivariate analyses. Finally, although we cannot assert that our findings are generalizable to SNV populations at large, we have no reason to suspect that our study inclusion criteria selected a subgroup with a distinct urinary tract cancer risk pattern.
In conclusion, the results of this study on urotoxic adverse events in CYC-treated SNV patients strengthen the link between urotoxic adverse events and cumulative CYC exposure and highlight oral CYC and perhaps also WG as additional independent risk factors. Our findings imply that hemorrhagic cystitis may have to be recognized as a marker of increased risk of urinary tract cancer, indicate the need for sustained heightened awareness when prescribing this drug to tobacco smokers, and support the suggested use of uroprotective measures, i.e., mesna and/or hyperhydration, during oral CYC treatment (2, 20). These results contribute to defining high-risk populations who should benefit from close and prolonged screening for urinary tract cancer and to preventing this rare but severe adverse event for patients receiving newly initiated CYC treatment for SNV.
- Top of page
- PATIENTS AND METHODS
- AUTHOR CONTRIBUTIONS
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Mahr had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Le Guenno, Mahr, Pagnoux, Dhote, Guillevin.
Acquisition of data. Le Guenno, Mahr, Pagnoux, Dhote, Guillevin.
Analysis and interpretation of data. Le Guenno, Mahr, Pagnoux, Dhote, Guillevin.