In our real-world study, the overall incidence of new-onset and aggravated hypertension during VSPi treatment was similar, at 24.3% and 25.2% respectively. These estimates are lower than reported in other observational studies and slightly higher than meta-estimates summarising the clinical trial experience where hypertension was measured [10–12, 23–25]. In clinical trial meta-analyses, the incidence of VSPi-induced hypertension (any grade) was 23.6% for bevacizumab [10], 23.1% for sorafenib [23], and 21.6% for sunitinib [11]. Cancer patients participating in randomized clinical trials are generally highly selected and in otherwise good health, and thus adverse events are observed less frequently than in real-world practice [26]. The two previous US claims-based studies, using International Classification of Diseases (ICD) codes for hypertension or codes for dispensed anti-hypertensive medicines [14, 15] observed higher incidence estimates, at 32% and 49% respectively. Our lower estimates may reflect the fact that we could only identify patients dispensed antihypertensives, not all those diagnosed with the condition. In one US real-world clinical practice study, 14.1% of patients with newly diagnosed hypertension during VSPi use were not treated with antihypertensives [14]. In a recent large pooled cohort study of Australian and New Zealand adults who were diagnosed with cancer during follow-up, 33% had untreated hypertension and 25% had treated hypertension at baseline [27]. Hypertension was ascertained via blood pressure measurements and treatment was based on self-report.
For patients with new-onset hypertension, the first antihypertensive dispensing was a median of 78 days from the index date, similar to a previous US claims-based analysis (96 days) [14]. The time interval to hypertension treatment differed by type of VSPi, and was significantly longer for patients treated with bevacizumab than oral TKIs. Among the oral TKIs, the shortest time interval was observed with pazopanib. VSPi-induced hypertension has been shown to develop rapidly, and to return to baseline after drug withdrawal [4, 28, 29]. Studies monitoring blood pressure have revealed blood pressure increases within hours to days after initiating oral TKIs [28, 29]. In one such study, the median time to first documented hypertensive response was 29 days in VSP-TKI treated patients [15].
For patients with new-onset VSPi-induced hypertension, the current guidelines recommend medications targeting the renin-angiotensin-aldosterone (RAAS) pathway such as ACE inhibitors, ARBs, and dihydropyridine CCBs as first-line therapy [30–32]. Most of the patients in our new-onset hypertension group were dispensed ACE inhibitors, ARBs, and CCBs and half started with monotherapy.
We identified patients with pre-existing hypertension based on the dispensing of an antihypertensive agent during the 6 months prior to starting VSPi treatment. Over 50% of our cohort met this criteria, consistent with US claims data [14]. These patients were dispensed an average of two antihypertensive medicines, with ACE inhibitors/ARBs most frequently dispensed. One quarter of our cohort who developed aggravated hypertension received dose intensification and these patients were dispensed an average of three antihypertensive agents after aggravation. Patients experiencing aggravated hypertension were predominantly dispensed ACE inhibitors, ARBs, and CCBs at baseline and their use of diuretics increased after aggravation. In US clinical data, a greater proportion of patients with pre-existing hypertension developed a hypertensive response (55% vs 40%) and pre-existing hypertension was a risk factor for VSPi-induced hypertension [15]. However, we found no significant difference between these two patient subgroups in our study, indicating an index of suspicion is warranted regardless of blood pressure history.
Consistent with US data [15], older age (≥ 60 years) and use of an oral TKI agent compared to bevacizumab were risk factors for new-onset hypertension after VSPi use. Only cancer type was associated with aggravation of hypertension in our cohort, with RCC patients displaying the greatest risk. This is consistent with prior evidence [11, 33, 34], and may be due to higher VEGF levels in RCC compared to other cancer patients, or it may be related to prior nephrectomy or reduced renal function [11, 34]. Alternatively, this may reflect information bias because we could only observe treated hypertension and those with RCC may be more likely to receive antihypertensive therapy.
The major limitation of our study is the lack of clinical information in the PBS database, noting that some antihypertensive medicines are prescribed for other indications. Thus, our data may over-estimate the incidence of clinically treated hypertension. We also could not describe cancer stage, related clinical and laboratory data, or detailed blood pressure measures, nor could we conduct risk factor analyses incorporating these factors. We did not have diagnoses data and ascertained hypertension through dispensing records. PBS records do not contain data for medicines dispensed to public hospital inpatients and we only observed anti-hypertensive medications dispensed in the community, so our study may over-estimate the time to first dispensed anti-hypertensive medication. Despite these limitations, our study covers the entire Australian population and both small molecule TKIs and antibody VSPi.
Our study provides real-world estimates of the incidence of, and risk factors for, VSPi-induced hypertension in a whole-of-population setting. Despite the well-known risks of cardiac adverse events associated with VSPi, there is little existing real-world evidence on hypertension incidence, risk factors and management in VSPi-treated patients. Our findings suggest that the real-world incidence of VSPi-induced hypertension is similar to that observed in pivotal clinical trials and they add valuable data for patients receiving care in routine clinical practice. Australian antihypertensive prescribing data thus appears to be a reliable and cost-effective proxy for clinically identified hypertension in this at-risk population.