Purpose To evaluate the relative efficacy of as needed versus treat and extend regimen for the treatment of neovascular age-related macular degeneration (AMD).
Methods We conducted a systematic review of studies that evaluated the efficacy of as needed or treat and extend regimen for neovascular AMD by searching multiple databases up to December 2013. Included studies were selected based on study duration of no less than 12 months, availability of outcome data, treatment protocol for as needed groups or pro re nata (PRN) receiving bevacizumab or ranibizumab, and all studies with treat extend protocols following the ‘inject and extend’ regimen. The outcome data were pooled and analysed.
Results 1046 peer reviewed articles meeting our initial search criteria were returned. After further review by two independent reviewers, 8 studies meeting treat and extend protocol and 62 studies meeting PRN protocol were included. The mean improvement in visual acuity in the PRN group was 5.4 ETDRS letters compared with 10.4 ETDRS letters in the treat and extend group. The PRN group received an average of 5.60 injections at 1 year compared with 8.09 in the treat and extend group. Central retinal thickness improved on average by 100.32 µ in the PRN group compared with 87.7 µ in the treat and extend group.
Conclusions Though our study suggests superiority of the treat and extend regimen to PRN treatment in a 12-month period, this review demonstrates the need for randomised clinical trials to confirm our findings and to evaluate long-term efficacy outcomes with these regimens compared with monthly therapy.
- Treatment Medical
Statistics from Altmetric.com
Age-related macular degeneration (AMD) is the leading cause of severe vision loss and blindness among Americans who are 65 years and older.
Recent studies have provided a better understanding of this complex multifactorial disease, and better equipped us with numerous options to treat this endemic problem. First-line therapy has evolved from laser photocoagulation and photodynamic therapy to more effective monthly antivascular endothelial growth factor (anti-VEGF) agents, which have been shown to prevent vision loss in the majority of patients, and to maintain and improve vision in many patients.1
However, the treatment burden of monthly injections on patients and physicians is significant and the socioeconomic impact on the healthcare system is immense.2 In addition, a review of the CATT study has shown that vision can be maintained with stable, chronic fluid, as seen on optical coherence tomography (OCT), and a dry macula may not indicate better long-term vision stabilisation. As such, lower frequency treatment regimens have been explored in studies in order to assess whether such regimens replicate visual and anatomical outcomes with monthly ranibizumab treatment regimens as used in the pivotal MARINA and ANCHOR trials. The PIER study employed a monthly loading dose for 3 months followed by quarterly injections3 while the PrONTO study used the same 3-month loading dose period, only to be followed by a monthly evaluation and decision to treat based on well defined criteria including visual acuity (VA) changes and the presence or absence of fluid on OCT.4 Both studies demonstrated improvements in vision compared with their controls, although PIER outcomes were less robust than MARINA or ANCHOR. The PrONTO study succeeded in reducing the number of injections required while maintaining VA but left the question of visit frequency and associated treatment burden unaddressed.
Despite these trials and the widespread use of ranibizumab in clinical practice, an ideal injection and monitoring frequency has not yet been established. A newer approach recommended by Spaide5 involves re-injecting patients monthly until resolution of macular fluid and then extending the interval between re-injection at subsequent visits provided that patients remain fluid-free on imaging. This inject and extend or treat and extend (T&E) regimen is gaining favour among treating physicians, even as many ongoing studies explore its efficacy.
While multiple studies have evaluated the as needed pro re nata (PRN) and T&E regimens with anti-VEGF agents for neovascular AMD and reported good outcomes, there is not any randomised clinical trial (RCT) yet in peer-reviewed literature that has reported a direct comparison between the two treatment modalities. To address this question, we performed a comprehensive systematic review to compare the available literature for efficacy outcomes associated with both approaches.
We conducted a systematic review of studies that evaluated the efficacy of PRN or T&E regimen for neovascular AMD.
Search strategy and inclusion criteria
The published literature was searched using strategies created by a medical librarian. Results were exported to EndNote for a total of 1046 unique citations. Authors of the original papers were not contacted to obtain further data and abstracts and unpublished studies were excluded. Included studies were selected based on study duration of no less than 12 months with the availability of outcome data. Studies with treatment protocol for the as needed groups following PrONTO criteria or PRN only with regular 4–6 weeks follow-up receiving bevacizumab or ranibizumab, and all studies with T&E protocols following Spaide5 ‘inject and extend’ strategy were included. Prospective and retrospective studies were included. (See online supplementary file for details of search strategy, exclusion criteria and risk of bias analysis.)
The following data were extracted from each trial: number of eyes, baseline VA, 12 month VA, baseline central retinal thickness (CRT) on OCT, 12 month CRT on OCT, number of injections administered over a 12 month period, and type of anti-VEGF agent used. The outcome data was pooled, and all VA was converted to ETDRS letter scores to analyse improved VA from baseline to 12 months after treatment, which was the primary outcome. We used Gregori et al6 conversion for Snellen vision to ETDRS letters. Secondary outcomes included number of injections administered over the 12-month period and the CRT change from baseline to month 12.
Each data set was recorded into a standard format for analysis, and then analysed separately, and appropriate summary data were derived for inclusion in each statistical analysis. Published data were included if appropriate. Outcome data was presented by weighted averages and mean data with SD. Subgroup analyses of PRN studies from RCTs were also presented. Mann-Whitney's non-parametrical test was used to compare statistical distributions. Spearman's correlation was also used to assess the relationship between change in VA and number of injections received in 1 year.
Literature search yielded 1046 peer reviewed articles meeting our initial search criteria. After further review of the articles, there were 8 studies meeting T&E protocol (n=1073 eyes) and 62 studies (n=10 716 eyes) meeting the PRN protocol (see online supplementary file for summarised data), of which 17 (n=2494 eyes) studies were RCTs. Seven T&E studies included in this review or 1035 eyes, provided all three outcome measures, including the change in VA (0–12 months), number of injections (over a 12 month period) and change in CRT on OCT (0–12 months) while one of the T&E studies including 38 eyes, evaluated two of the three outcome measures: the change in VA (0–12 months), and the number of injections (over the 12-month period). For the PRN studies, 46 in this review including 7913 eyes, provided all three outcome measures, including the change in VA (0–12 months), number of injections (over a 12 month period) and change in CRT on OCT (0–12 months) while 16 of the PRN studies with 2803 eyes, evaluated two of the three outcome measures, including the change in VA (0–12 months) and number of injections (over the 12-month period) (figure 1).
Risk of bias
Study bias was significantly dependent upon the study design with lower risk of selection bias in the PRN studies that were RCTs, compared with the prospective and retrospective studies. This was also true for performance, and detection bias, as there was no masking in non-randomised controlled trials. In comparison, attrition bias was lower in the retrospective studies compared with the prospective and RCTs, as incomplete data was not assessed in the retrospective studies. Raw data was not collected from the RCTs, leading to the possibility of selective reporting, but review of the studies demonstrated low risks in this area, compared with the non-randomised studies where the risk of selective reporting was higher. Overall, low risk was noted in the RCTs, and a range of low–medium risk for the prospective and retrospective studies. The evidence table highlights the areas of risk analysis (see online supplementary file for evidence table).
Primary and secondary outcomes
The improvement in VA at 1 year in the PRN group had a weighted mean of 3.5 or a mean of 5.4 (n=10 716, SD: 4.5, CI 4.28 to 6.43) ETDRS letters compared with a weighted mean of 9.17 or a mean of 10.4 (n=1073, SD: 3.8, CI 7.4 to 13) in the T&E group.
In the PRN group, the average number of injections received was a weighted mean of 5.3 or a mean of 5.6 (n=10 716, SD: 0.99, CI 5.32 to 5.79) at 1 year compared with a weighted mean of 8.34 or a mean of 8.09 (n=1.073, SD: 0.66, CI 7.6 to 8.6) in the T&E group.
CRT improved on average by a weighted mean of 99.2 or a mean of 100.32 (n=7913, SD=40.69, CI 89.1 to 111.54) µ in the PRN group compared with a weighted mean of 85.71 or a mean of 87.7 (n=1035, SD=27.67, CI 64.58 to 110.8) µ in the T&E group.
A subgroup analysis of the PRN studies from RCTs demonstrated an improvement of VA at 1 year by a weighted mean of 5.08 or a mean of 5.8 (n=2494, SD=2.7, CI 4.4 to 7.2) ETDRS letters. The average number of injections received in this subgroup was a weighted mean of 6.6 or a mean of 6.5 (n=2494, SD =0.88, CI 6 to 6.9) at 1 year and CRT improved on average by a weighted mean of 122.2 or a mean of 110 (n=2494, SD=32, CI 90 to 120) µ.
A statistically significant difference in change in VA and number of injections was found between the T&E and all PRN studies (p=0.0006 and p<0.0001, respectively) and also between the T&E and RCT PRN subgroup (p=0.0051 and p=0.0002, respectively). The change in CRT between treatment regimens was not found to have a statistically significant difference between T&E versus all PRN (p=0.41) and T&E versus RCT PRN subgroup (p=0.1784) (table 1).
A positive correlation between having more injections within a year and having greater mean gain in ETDRS letters was demonstrated by Spearman's correlation (r=0.412, p=0.0002).
In light of the treatment burden of monthly anti-VEGF injections and the potential risk of ocular severe adverse events, an approach that reduces the demands of therapy while maintaining good visual outcomes is desirable. This study represents a systematic review of two popular treatment regimens (T&E vs PRN) for neovascular AMD that can potentially reduce some of the burden on the healthcare system. In order to assess and compare these regimens we conducted this review. Usual meta-analysis was not feasible in the absence of RCTs assessing the T&E regimen. However, collection and analysis of available studies with pooling of the data that meet inclusion criteria was performed and weighted means and analysis were completed. Individual analysis between drug types was not done, and as such, ranibizumab and bevacizumab were considered to be equivalent in the studies.
In addition, subgroup analysis of PRN data from RCTs was done in order to isolate and assess the most evidence-based representative data for that regimen. The subset results were similar to that from the overall analysis of the included PRN studies, as seen in table 1. Therefore, we decided to include all study designs, as we felt that many of the other studies, some of which included highly referenced prospective series (PrONTO study), would represent a more realistic or ‘real world’ comparison. Also, analysis of outcomes from many of these studies demonstrated comparable results, highlighting the fact that well conducted studies can provide evidence for determining a treatment regimen.
Haller, in her literature review, noted evidence suggesting that either a treat-as-needed or, possibly, a treat-and-extend regimen provided a reasonable alternative to monthly injections of bevacizumab or ranibizumab.7 Our study supports the benefit of both regimens and suggests the superiority of T&E regimen to PRN treatment over a 12-month period. The difference in the primary outcome—the change in VA by ETDRS letters gained—was statistically significant with greater gain in the T&E group (10.4) versus the PRN group (5.4), which is similar to the results obtained from the only published trial comparing the two regimens by Oubraham et al.8 This retrospective non-randomised case series found a mean gain of 10.8 letters in the T&E group versus 2.3 in the PRN group. In this study, the PRN results were slightly lower than in our study, which might be attributed to the lack of consistent monthly follow-up (8.8 visits instead of 12 visits in 1 year) that occurred despite the planned monthly evaluation in their protocol. Overall, we felt that including the entire spectrum of the literature in this analysis helped to portray a more inclusive and comprehensive outcome, but also adds significant bias that may take away the validity of our results and statistical analysis. This however, supports the notion that RCTs comparing the two treatment regimens are vital to answering this question.
From this study, we also found a correlation between number of injections received in 1 year and the mean gain in ETDRS letters. This might explain why the T&E group had a larger gain in ETDRS letters compared with the PRN group that received fewer injections over the year. This theory has previously been proposed, but no analysis or correlation was assessed as in our paper. Another plausible explanation for patients doing better in the T&E group is that unlike in the PRN group, there is a mandatory treatment at every visit, which could minimise intervening episodes of recurrent exudation and resultant long-term visual compromise.
It is also important to note that this study does not discredit the PRN regimen. Our results indicate improvement with either regimen. In addition, the inferior outcomes in the PRN studies could be due to patients not receiving treatment when indicated. Furthermore, missed treatment opportunities might also be represented by patients missing routine monthly visits or by diagnostic failures linked to the lower sensitivity of the previous generations of OCT. This might be addressed in repeat analyses and future RCTs employing the as needed regimen with newer diagnostic devices.
The greater interval improvement in CRT in the PRN studies over the T&E studies was not expected, though this difference did not attain statistical significance. In addition, not all the studies included had OCT data reported and, as such, resulted in a limited sample size. Many studies have also demonstrated the lack of correlation between CRT and VA outcome.9 ,10 These data suggest that improvement in VA cannot be entirely attributed to change in CRT though the two are often intimately associated.
Patients in our study should be representative of the majority of patients seen in clinical practice and included studies with a wide range of baseline VA, with treatment-naïve and prior treatment-responders, and with those employing either ranibizumab or bevacizumab. On the last point, both medications were found to be equivalent when using the same treatment regimen as noted by the CATT trial.11 Only the IVAN trial that was included, presented PRN data for patients using either ranibizumab or bevacizumab. This study was included as there was no switching between medications (ranibizumab or bevacizumab due to treatment-desensitisation).12
Among the limitations of this study is the lack of consistent control groups to perform a standard meta-analysis and the limited number of studies evaluating the T&E regimen. However, it is still becoming the modality of choice as noted by the American Society of Retina Specialists preferences and trends survey. Another limitation is that while the large number of subjects involved by pooling data from various studies raises the possibility of additional significant associations, the data are confounded by a lack of overall randomisation. This underscores the fact that the results only point to an association and not necessarily causation. The implication is that this study cannot conclusively indicate the superiority of any regimen. Rather, it demonstrates the likelihood of greater mean ETDRS letters in a T&E regimen over an as needed regimen and highlights the need for future prospective RCTs evaluating the efficacy of T&E regimen versus PRN regimen and monthly injections.
Dr Mae Gordon (Department of Ophthalmology, Washington University, St Louis, Missouri, USA) assisted and provided input for the article reviews and statistical analysis.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Data supplement 1 - Online supplement
Contributors Design and conduct of the study: DC-Y and RSA; collection: DC-Y, RSA, TE, AH and SF: management: DC-Y, RSA and TE: analysis: DC-Y, RSA and TE; interpretation of the data: DC-Y, RSA and TE; and preparation, review or approval of the manuscript: DC-Y, RSA and TE.
Competing interests RSA: Founder: Metro Midwest (Birmingham, Michigan, USA), Seven Sights (St Louis, Missouri, USA) Consultant: Genentech (San Francisco, California, USA), Regeneron (Tarrytown, New York, USA), Novartis (Basel, Switzerland), Bayer (Leverkusen, Germany), Valeant (Quebec, Canada), Bausch and Lomb (Rochester, New York, USA), Allergan (Irvine, California, USA), Alimera (Alpharetta, Georgia, USA) Speaker: Genentech (San Francisco, California, USA), Regeneron (Tarrytown, New York, USA), Novartis (Basel, Switzerland), Bayer (Leverkusen, Germany) Equity: Ophthotech (New York, New York, USA).
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.