Intravitreous bevacizumab and blood pressure: does ‘safe’ mean ‘safe enough’?


Focke Ziemssen MD
University Eye Hospital Tübingen
Schleichstrasse 12
72076 Tübingen
Tel: + 49 7071 29 84828
Fax: + 49 7071 29 5143


We read with great interest the communication of Kernt et al. (2007) describing their investigation into the behaviour of intraocular pressure (IOP) and blood pressure (BP) following intravitreal bevacizumab injection. Although we very much appreciate the special care taken in the safety assessment of this off-label treatment, we can not relate to the general statements because of the methodological limitations of the study.

Measuring BP is not as easy as eye specialists might suggest and judging whether or not BP is ‘sufficiently’ controlled is not a trivial decision. Thus, we want to raise some questions, more to be sure of making the right treatment decisions in our own hypertensive patients than to aggrandize by sophisticated criticism.

(1) The authors mention the high comorbidity of age-related macular degeneration and hypertension, both of which show higher incidences, and even exponentially increase, with age. We totally agree that control of BP is of the utmost relevance. The seventh report of the Joint National Committee (JNC 7) on Prevention, Detection, Evaluation and Treatment of High Blood Pressure issued new guidelines for the definition and management of hypertension (Chobanian et al. 2003). According to this report: ‘The relationship between BP and risk of cardiovascular disease events is continuous, consistent, and independent of other risk factors. The higher the BP, the greater the chance of myocardial infarction, heart failure, stroke and kidney disease.’ Given this close and direct relationship, we wonder whether our colleagues made any concrete recommendations to their patients. How many individual patients showed BP values – whether affected or not by the injection and the drug bevacizumab – which signalled a need to intensify antihypertensive medication?

From our point of view, it might not be wise to look for a significant change in BP over the time−course of the study period only, and to leave some patients with insufficiently controlled BP. Boxplots and medians point out that many patients may suffer from insufficient pharmacotherapy (the target level for uncomplicated hypertension is 140/90 mmHg, and 130/80 mmHg in cases of diabetes or nephropathy). Despite several obstacles in the real world, we would like to encourage all ophthalmologists to think in terms of risk-based management in hypertension and to make further efforts while trying to maintain an interdisciplinary approach (Scheltens et al. 2007).

(2) Blood pressure shows high inter- and intraindividual variability as well as diurnal oscillation (Ceyhan et al. 2003). Unfortunately, the paper does not contain any specification of the statistical tests applied to cope with this variance. Was the sample of 45 patients sufficient in number to show normal distribution (especially given the outliers at the 6-hour measurement)? Was the intraindividual information evaluated by using matched-pair analysis? Were the injections (and the measurements) performed at particular times of day?

We understand that Kernt et al. (2007) did not want to overlook early changes in BP after the injection. However, given the potential white-coat effect, would ambulatory monitoring not be superior to single measurements taken at 1, 3 and 6 hours? Clinical sphygmomanometric measurement of BP is known to show low short- and longterm reproducibility (Fotherby & Potter 1993). Waiting for 6 hours or returning several times to the outpatient clinic might aggravate the artificial monitoring situation and therefore limits further conclusions for other patients, who may be discharged from the clinic and go on to pursue other activities after the injection.

(3) Because of the decline in diastolic BP, Kernt et al. (2007) hypothesize a relevant influence of conceivable ‘surgery-related stress’ on preoperative BP, which they refer to as ‘baseline’. The exact time interval between the measurement ‘before’ and the injection is not given in the text. Previous works have shown that the stress-induced increase in sympathicotonus can vary a lot between individuals. In a subgroup of patients, the assumed stable BP (no significant change in time−course) could therefore mean simply a persistently increased BP compared with the real, stress-free situation before the injection. If the first measurements, to which all the statistics were related, were not reliable, the conclusion would appear to be quite adventurous. By claiming that emotional factors are dependent on the procedure, the authors admit that their baseline does not constitute a consistent reference.

Assessing the risk profile of anti-VEGF (vascular endothelial growth factor) drugs may be very important. Already a slight increase in BP implies a direct and relevant rise in cardiovascular events and mortality. We know that large, industry-driven safety studies, previously performed and taken into account during the approval process, were also restricted to incidental and clinical measurements. However, the designs of such studies may have been influenced by the financial interests of the manufacturer and were therefore often not in accordance with clinical guidelines.

From our point of view, increased awareness will be necessary for at least as long as clinical data indicate a tendency towards a dose-dependent risk of stroke after treatment with the closely related drug, ranibizumab (Rosenfeld et al. 2006).