Bevacizumab and ranibizumab tachyphylaxis in the treatment of choroidal neovascularisation
- Julie L Gasperini1,
- Amani A Fawzi2,
- Ani Khondkaryan2,
- Linda Lam2,
- Lawrence P Chong3,
- Dean Eliott4,
- Alexander C Walsh2,
- John Hwang2,
- SriniVas R Sadda2
- 1South Coast Retina Center, Torrance, California, USA
- 2Doheny Eye Institute, Doheny Retina Institute, Department of Ophthalmology, Keck School of Medicine of the University of Southern California, Los Angeles, California, USA
- 3VMR Institute, Huntington Beach, California, USA
- 4Massachusetts Eye and Ear Infirmary, Boston, Massachusetts, USA
- Correspondence to Dr SriniVas Sadda, Doheny Eye Institute, 1450 San Pablo Street, DEI 3623, Los Angeles, CA 90033, USA;
- Accepted 27 June 2011
- Published Online First 26 July 2011
Aims To evaluate the effect of switching to bevacizumab or ranibizumab after developing tachyphylaxis during anti-vascular endothelial growth factor (VEGF) therapy for choroidal neovascularisation (CNV).
Methods The authors reviewed the records of all patients who received both ranibizumab and bevacizumab for treatment of CNV to identify those who developed tachyphylaxis, defined as optical coherence tomography evidence of initial decreased exudation followed by lack of further reduction or an increase in exudation. Signs of exudation included subreitnal fluid (SRF), pigment epithelial detachment (PED) and/or cystoid macular oedema (CMO).
Results 26 eyes were included. 10 were initially treated with bevacizumab and then changed to ranibizumab for persistent SRF, PED and/or CMO. Of these, seven had occult CNV and three had predominantly classic CNV. One eye in the occult CNV group did not respond after being switched to ranibizumab. Six eyes had a positive therapeutic response, after one injection in four eyes, and after two or three injections in one eye each. In the classic group, two responded to ranibizumab and one did not. Sixteen eyes were initially treated with ranibizumab before changing to bevacizumab. Of these, 15 had occult CNV and 1 was predominantly classic. Three of the 16 eyes failed to respond to bevacizumab; 6 improved after one injection and 5 after two injections.
Conclusions Patients with CNV who develop tachyphylaxis to ranibizumab or bevacizumab may respond to another anti-VEGF drug. The majority of cases (81%) in this series demonstrated at least some response after switching therapies.
Age-related macular degeneration (AMD) is the leading cause of blindness in individuals over 55 years of age in the developed world.1 The development of choroidal neovascularisation (CNV) occurs in the exudative form of AMD, which is associated with severe vision loss.2 In recent years, the treatment for exudative AMD has expanded to include intravitreal pharmacological agents such as ranibizumab (Lucentis; Genentech Inc., South San Francisco, California, USA) and bevacizumab (Avastin; Genentech, Inc., South San Francisco, California, USA). These agents have been found to prevent the loss of vision or even improve visual acuity in many patients.3 4 Both drugs neutralise all known active isoforms of vascular endothelial growth factor (VEGF), which plays a central role in the development of CNV. Ranibizumab is a recombinant humanised monoclonal antibody fragment that is currently FDA approved for the treatment of exudative AMD.3 Bevacizumab is a recombinant humanised monoclonal antibody and is used off-label by many practitioners to treat CNV4 and other VEGF-mediated diseases, such as central retinal vein occlusion5 6 and proliferative diabetic retinopathy.7–9 Intravitreal anti-VEGF therapy is usually not durable; thus, these medications typically must be administered chronically and repeatedly to maintain their effect.10
Long-term efficacy of a drug can be affected by tachyphylaxis, which is defined as a decreasing therapeutic response to a pharmacological agent following repeated administration over time.11 Tachyphylaxis has been reported to occur with drugs such as salmeterol,12 brimonidine13 and infliximab.14 Keane et al first suggested possible tachyphylaxis in the treatment of neovascular AMD after investigating retinal morphology by optical coherence tomography (OCT) following treatment with ranibizumab for CNV.15 Their report concluded that neurosensory retinal oedema and subretinal fluid (SRF) showed an early reduction to nadir after the initiation of ranibizumab therapy; however, the effect on the retina was attenuated over time, suggesting possible tachyphylaxis. A study by Schaal et al followed OCT volumetric changes after the administration of intravitreal bevacizumab for the treatment of exudative AMD.16–19 The standardised volumetric change index decreased over time, indicating a possible tachyphylactic response to bevacizumab. A report by Forooghian et al identified six eyes that developed tachyphylaxis after repeated treatment with intravitreal bevacizumab at the National Eye Institute.20
The purpose of this study was to determine whether patients with CNV who had an initial morphological response to ranibizumab or bevicizumab therapy but later developed an attenuated response despite repeated injections would show an improved response if their treatment regimen was changed to a different anti-VEGF drug.
We performed a retrospective chart review of patients treated for CNV at the Doheny Eye Institute from September 2006 to April 2009 who had received both intravitreal ranibizumab and bevacizumab and who had OCT volume scans of the macula at each visit. Approval for data collection and analysis was obtained from the Institutional Review Board at the University of Southern California and was conducted according to the principles set forth in the Declaration of Helsinki.
Evidence of exudation was defined as the presence of SRF, pigment epithelial detachment (PED) and cystoid macular oedema (CMO) on OCT scans. The presence and extent of these features of exudation were evaluated at each visit for each case by inspecting the individual B-scans of the volume acquisition. Because automated measurements of these various individual fluid compartments (eg, volume of SRF) are not provided by the OCT instrument, differences in the extent of these features were judged qualitatively by simultaneous comparison of B-scans (from approximately the same retinal location) between visits. Although this qualitative method was probably not sensitive enough to detect subtle changes, definite reduction or increase in exudation in any of these compartments could be readily identified. A definite (ie, qualitatively visible) reduction in at least one of these compartments (neurosensory retina, subretinal space or PED) with no increase in fluid in any compartment was deemed evidence of a positive therapeutic response. A lack of definite reduction or an increase in exudation in any compartment was considered evidence of poor response. For inclusion in the study, eyes had to demonstrate evidence of an initial positive morphological response (reduction in exudation) to treatment with intravitreal injections of either ranibizumab or bevacizumab, followed by an eventual poor response to treatment, despite multiple (at least two) monthly injections. In addition, to be included, these poorly responding eyes must have been switched to an alternate anti-VEGF drug, either bevacizumab or ranibizumab. Patients were excluded from the study for the following reasons: (A) if they did not meet the inclusion criteria, (B) if the treatment was changed based on physician or patient preference (for reasons other than tachyphylaxis), (C) if the patient did not respond to the initial treatment (ie, treatment failures) or (D) if the patient missed a scheduled monthly visit and treatment and thus had a resultant increase in exudation which could not be ascribed to tachyphylaxis.
A comprehensive eye exam and OCT imaging were performed at each visit, and angiography was obtained at baseline to confirm the diagnosis and type of CNV. A StratusOCT (Carl Zeiss Meditec, Dublin, California, USA), Cirrus HD-OCT (Carl Zeiss Meditech) or Topcon 3D-OCT (Topcon Medical Systems, Paramus, New Jersey, USA) was used to qualitatively assess the presence of CMO, SRF and PED. Because this was a retrospective study, the drug used for the initial treatment and the subsequent retreatment strategy for all patients and visits was not standardised. The administered dose for the intravitreal drugs was 1.25 mg (0.05 ml) of bevacizumab or 0.5 mg (0.05 ml) of ranibizumab. Patients were treated at the discretion of the ophthalmologist, with an OCT-guided retreatment protocol similar to that described in the PrONTO study.21 In general, eyes were treated if one or more features of exudation were present on OCT, and treatment was deferred if the macula was free of SRF, PED and CMO.
Data collected for each patient included age, sex, the eye requiring treatment, best-corrected Snellen visual acuity, angiographic CNV lesion classification at the time of initial intravitreal injection, interval between treatments, number of treatments with ranibizumab and bevacizumab, and response to treatment, including OCT findings.
A total of 39 charts were reviewed. Twenty-six eyes of 25 patients met the inclusion criteria for the study. Of the 25 patients in the study, 7 (28%) were men and 18 (72%) were women. The average age of the patients was 80 years (range 65–98). Twenty-five eyes were treated for typical CNV secondary to AMD and one eye had CNV with features suggestive of occult CNV and idiopathic polypoidal choroidal vasculopathy (IPCV). The interval between appointments and treatments was on average 4 weeks for patients being treated with ranibizumab and 6 weeks for patients being treated with bevacizumab. The average length of follow-up was 13 months (range 6–28).
Eyes initially treated with bevacizumab
Ten (38%) of the 26 eyes were initially treated with bevacizumab for CNV secondary to exudative AMD (table 1). These eyes received on average seven injections (range 3–13 injections) of bevacizumab before the treatment regimen was changed to ranibizumab for apparent tachypylaxis (ie, attenuated response) and presence of persistent or worsening SRF, PED and/or CMO. Three of the 10 eyes were initially treated with bevacizumab for predominantly classic CNV and seven eyes for an occult with no classic CNV, of which six eyes had a PED on OCT.
After an initial morphological response to bevacizumab (decreased PED, SRF and/or CMO), the response attenuated over time. Treatment was switched to ranibizumab for increasing PED and SRF (four eyes), persistent SRF (four eyes) and increasing SRF (one eye), or increasing SRF and CMO (one eye). After the regimen was switched to ranibizumab, one eye failed to respond due to worsening SRF, while two of the three eyes with classic CNV had complete resolution of SRF. In these patients, the SRF resolved after one injection in one eye and after four injections in a second eye. Of the seven eyes with occult CNV, four responded after one ranibizumab injection, one after two injections and one after three injections. Two of the seven eyes achieved complete resolution of SRF. Of these, one eye had complete resolution after two ranibizumab injections and another after three injections (figure 1), but treatment was continued for persistent PED in both eyes. Treatment was continued in the other three eyes with occult CNV for improved but persistent SRF and PED. One of seven eyes with occult CNV failed to respond to the switch to ranibizumab. After the change in the treatment regimen, eyes received, on average, seven injections of ranibizumab (range 1–16 injections).
For example, patient no. 6 who received 10 initial bevacizumab injections for a large PED and SRF had a positive response after the first bevacizumab injection with decreasing PED and SRF. There was a continued reduction in the size of the PED and the amount of SRF with subsequent injections. However, after the eighth bevacizumab injection there was no further improvement in the size of the PED or amount of SRF, so after the tenth bevacizumab injection, the therapy was changed to ranibizumab. After the first ranibizumab injection, there was a positive response with decreasing PED and SRF with continued improvement but not complete resolution after the sixth ranibizumab injection.
Eyes initially treated with ranibizumab
Sixteen of the 26 (62%) eyes were initially treated with ranibizumab (table 2). Fifteen of the eyes were being treated for typical CNV secondary to AMD, and one eye had features of IPCV. Of the 16 eyes initially treated with ranibizumab, one eye was treated for classic CNV and 15 eyes for occult CNV (13/15 eyes had a PED on OCT). After an initial positive morphological response to ranibizumab, these eyes developed apparent tachyphylaxis, after, on average, five injections of ranibizumab (range 2–11 injections). Tachyphylaxis manifested as increasing PED and SRF (two eyes), persistent PED and SRF (four eyes), increasing PED with SRF (one eye), increasing SRF with PED (one eye), increasing SRF (three eyes), persistent SRF (three eyes), increasing CMO (one eye), and increasing CMO with PED (one eye).
Of the 16 eyes, three (19%) failed to respond after the change to bevacizumab. All three of these eyes had occult CNV with SRF and PED. Of the remaining 13 (81%) eyes, 8 (62%) showed a response (reduction or resolution of exudation) after a single injection of bevacizumab and 5 (38%) after two injections (figure 2). Six of the 15 eyes treated for occult CNV with PED had complete resolution of SRF (after one injection in two eyes and after two injections in four eyes), and two of these eyes also had complete resolution of the PED (after two injections in one eye and after three injections in the second eye; figure 3) while four eyes had persistent PED despite continued treatment. Treatment was ongoing in the remaining 10 eyes for improving SRF, PED and/or CMO. On average, three injections of bevacizumab (range 1–6 injections) were administered.
As an example, patient no. 23 had two initial ranibizumab injections. After the first injection, the size of the PED decreased, the SRF resolved and vision improved from 20/100 to 20/30. After the second injection, vision decreased to 20/40, the PED increased in size and there was a recurrence of SRF and CMO, so the therapy was changed to bevacizumab. After the first bevacizumab injection, there was a positive response with decreased PED and SRF. After the third injection, there was resolution of CMO and SRF but a persistent PED.
In this retrospective review, we observed that switching to a different anti-VEGF agent was associated with improved response in 21 (81%) of the 26 eyes that developed tachyphylaxis. The improved response was found in patients initially treated with either ranibizumab (13/16 eyes, 81%) or bevacizumab (8/10 eyes, 80%). Of the five eyes that failed to respond, treatment was initiated with bevacizumab for classic CNV in one eye and for occult CNV in one eye, and with ranibizumab for occult CNV in three eyes. Of the 21 eyes that responded, 14 eyes responded after one injection, six eyes after two injections and one eye after three injections. At the last follow-up, 10/21 eyes had complete resolution of fluid (eight eyes with occult CNV and two eyes with classic CNV), while the remaining 11 eyes were being treated for improved but persistent exudation.
Almost all eyes had typical CNV secondary to AMD, while one eye had occult CNV with features of IPCV. We did not find that one particular CNV lesion type was more prone to developing an attenuated or diminished therapeutic response over time. The majority of eyes were being treated for occult CNV (22 of 26 eyes, 85%), while four eyes had classic CNV. Most eyes with occult CNV had evidence of PED on OCT (19 eyes), and of those 19 eyes, only 2 had complete resolution of the PED at the last follow-up visit. Previous studies have also observed that PEDs tend to regress more slowly than SRF or CMO.4 15 21
The attenuated therapeutic response over time was observed to occur whether treatment was initiated with ranibizumab or bevacizumab. This decrease in the biological response to these intravitreal anti-VEGF drugs may be attributed to the phenomenon of tachyphylaxis. Other reports have also suggested tachyphylaxis to ranibizumab or bevacizumab.15 16 20 By definition, the attenuated response occurs after repeated administration of a drug over time. We observed that some patients developed a decreased response quickly, after two anti-VEGF injections, while others did not develop tachyphylaxis until after 10 or 11 injections. In the report by Forooghian et al, the median time to develop tachyphylaxis was 100 weeks, and the median number of intravitreal bevacizumab treatments prior to developing tachyphylaxis was eight.20 Additionally, Schaal et al reported that approximately three injections were required before the efficacy decreased to 50% of the initial OCT response.16
Tachyphylaxis has been described with other pharmacological agents.12–14 In general, the mechanisms of drug tolerance include metabolic and cellular tolerance.11 Metabolic tolerance is the result of alterations in drug absorption, distribution or metabolism, which decreases the effective concentration of the drug. With cellular tolerance, the response to the drug is decreased by cellular mechanisms. For example, receptors may adapt to the continued presence of the drug, by reducing the number of receptors available to the drug or by reducing their sensitivity to the drug. The pattern of tachyphylaxis varies according to several parameters, such as administration frequency, dose, receptor expression patterns on the target and the presence of antagonists.12–14 16 There are also other possible causes for a decrease in biological response. For example, an increase in other angiogenic signalling pathways may compensate for the blocked activity of VEGF and the response to anti-angiogenic treatment may be dependent on the maturity of the vessels.22 Other specific systemic and local causes of tachyphylaxis to intravitreal anti-VEGF drugs have been suggested. Macrophages located within the CNV tissue may respond to VEGF blockade by upregulating the production of VEGF.20 Systemic circulating antibodies to bevacizumab23 and ranibizumab3 may neutralise the drugs' effect. Certain ocular/host factors have been proposed to explain the attenuated therapeutic response such as a change in CNV lesion type, loss of retinal pigment epithelial function or the development of chronic inflammatory changes.20
The pharmacological literature suggests that tachyphylaxis can be reduced or avoided through drug holidays or drug-free intervals, increases in the drug dosage, combination therapy using drugs with different mechanisms of action or the use of another drug from the same class.11 The present study is an example of a fourth option since bevacizumab, a full-length humanised monoclonal antibody, and ranibizumab, a humanised antigen binding fragment, have a common molecular lineage. Both drugs are proteins that were genetically modified from the same murine monoclonal antibody against VEGF. In the case series by Forooghian et al, a higher dose of bevacizumab was used in five eyes that developed tachyphylaxis.20 However, none of these eyes demonstrated a positive therapeutic response. Of their patients, one experienced a resolution of fluid after the dose was increased to 2.5 mg but subsequently developed tachyphylaxis with repeated administration at the higher dose.20 The report by Schaal et al suggested that combined pharmacotherapy with triamcinolone acetate lessened the effect of tachyphylaxis with intravitreal bevacizumab.16
Our study has several limitations, including in particular its retrospective design and relatively small sample size. In addition, there were multiple treating physicians without a mandated, standardised retreatment protocol, although all the physicians belonged to the same group practice and had similar practice patterns. Moreover, the OCT features of exudation were evaluated qualitatively rather than by quantitative measurements, thus limiting the sensitivity to detect subtle changes. Finally, in our study intravitreal injections were administered on an as-needed basis; the therapeutic response may have been sustained if the treatment had been administered on a regimented basis of every 4 weeks for ranibizumab and every 6 weeks for bevacizumab.15
Nonetheless, despite these limitations, the findings from this study corroborate previous reports that eyes being treated for CNV with either ranibizumab or bevacizumab can develop a diminished therapeutic response over time. This attenuated response may be attributed to the phenomenon of tachyphylaxis. Importantly, our study suggests that the majority of these patients may respond favourably to a change in the treatment regimen to another anti-VEGF drug. Patients may, however, require multiple injections to demonstrate a favourable treatment response and reversal of the tachyphylaxis effect. These preliminary findings should be evaluated in future prospective trials to more fully elucidate the mechanisms of tachyphylaxis in anti-VEGF therapy and strategies to combat it.
Linked articles 203893.
Presented in part at the Association for Research in Vision and Ophthalmology Annual Meeting, Fort Lauderdale, FL, April 2008 and American Society of Retina Specialists Meeting, Canada, 2010.
Funding Supported in part by an unrestricted grant to the Department of Ophthalmology from Research to Prevent Blindness Inc., New York, NY, and a grant from the Fletcher Jones Foundation.
Competing interests LPC: Allergan Consultant; DE: Genentech Speaker; ACW: co-inventor of Doheny intellectual property related to optical coherence tomography that has been licensed by Topcon Medical Systems and receives research support from Carl Zeiss Meditec and Optovue, Inc.; SVRS: co-inventor of Doheny intellectual property related to optical coherence tomography that has been licensed by Topcon Medical Systems, consultant for Heidelberg Engineering, Genentech and Allergan, and receives research support from Carl Zeiss Meditec and Optovue, Inc.
Provenance and peer review Not commissioned; externally peer reviewed.