Fibromuscular dysplasia (FMD) encompasses a heterogeneous

Transcription

1 Efficacy of Revascularization For Renal Artery Stenosis Caused by Fibromuscular Dysplasia A Systematic Review and Meta-Analysis Ludovic Trinquart, Claire Mounier-Vehier, Marc Sapoval, Nathalie Gagnon, Pierre-François Plouin Abstract In patients with fibromuscular dysplasia and renal artery stenosis, renal artery revascularization has been used to cure hypertension or to improve blood pressure control. To provide an up-to-date assessment of the benefits and risks associated with revascularization in this condition, we performed a systematic review of studies in which hypertensive patients with fibromuscular dysplasia renal artery stenosis underwent percutaneous transluminal renal angioplasty or surgical reconstruction. We assessed how often periprocedural complications and hypertension cure and improvement occurred. We selected 47 angioplasty studies (1616 patients) and 23 surgery studies (14 patients). Combined rates of hypertension cure, defined according to the criteria in each study, after angioplasty or surgery were estimated to be 46% (95% CI: 4% to 52%) and 58% (95% CI: 53% to 62%), respectively, with substantial variations across studies. The probability of being cured was negatively associated with patient age and time of publication. Cure rates using current definitions of hypertension cure (blood pressure 14/9 mm Hg without treatment) were only 36% and 54% after angioplasty and surgery, respectively. The combined risks of periprocedural complications were 12% and 17% after angioplasty and surgery, respectively, with less major complications after angioplasty than surgery (6% versus 15%). In conclusion, angioplasty or surgical revascularization yielded moderate benefits in patients with fibromuscular dysplasia renal artery stenosis, with substantial variation across studies. The blood pressure outcome was strongly influenced by patient age. (Hypertension. ;56: ) Key Words: renal artery stenosis fibromuscular dysplasia revascularization percutaneous transluminal angioplasty surgical reconstruction systematic review meta-analysis Fibromuscular dysplasia (FMD) encompasses a heterogeneous group of idiopathic, segmental, nonatherosclerotic diseases of the musculature of arterial walls, leading to the narrowing of small and medium-sized arteries. 1 FMD is the second most frequent cause of renal artery stenosis (RAS) after atherosclerosis. Renal artery revascularization through surgical reconstruction and percutaneous renal transluminal angioplasty (PTRA) has been used to treat hypertensive patients with RAS and presumed renovascular hypertension. Atherosclerotic RAS mostly affects elderly patients with frequent renal parenchymal disease. 2 Conversely, FMD predominantly affects women in their 3s or 4s with normal kidney function. Because advanced age and chronic kidney disease are associated with a poor outcome after etiologic treatment, 2,3 blood pressure (BP) outcome after renal artery revascularization is expected to be more favorable in patients with FMD than in patients with atherosclerosis. An overview, published within years of PTRA being first used, reported a mean rate of hypertension cure after PTRA (defined according to each study criteria) of 5% in 193 patients with FMD RAS versus 19% in 464 patients with atherosclerotic RAS. 4 Over the last years, several randomized, controlled trials and systematic reviews have assessed the efficacy of surgical reconstruction or PTRA for atherosclerotic RAS and confirmed the limited benefit of revascularization in this condition. 5 9 By contrast, there have been no randomized, controlled trials and no comprehensive systematic reviews assessing the BP outcome associated with revascularizing FMD stenoses. Our primary objective was to systematically review and assess the rate of hypertension cure after renal artery revascularization (using surgery or PTRA) in patients with hypertension and fibromuscular RAS. Secondary objectives were to assess the technical success rates of revascularization and Received March 5, ; first decision March 27, ; revision accepted June 21,. From the Institut National de la Santé et de la Recherche Médicale Centre d Investigation Clinique et Epidémiologie Clinique 4 (L.T.), Paris, France; Université Lille 2 (C.M.-V.), Faculté demédecine Henri Warembourg, Centre Hospitalier Universitaire et Régional de Tours, Hôpital Cardiologique, Service de Médecine Vasculaire et Hypertension, Lille, France; Université Paris Descartes (M.S.), Assistance Publique-Hôpitaux de Paris, Hôpital Européen Georges Pompidou, Service de Radiologie Cardiovasculaire, Paris, France; Université Paris Descartes (N.G., P.-F.P.), Assistance Publique-Hôpitaux de Paris, Hôpital Européen Georges Pompidou, Service de Médecine Vasculaire et Hypertension, Institut National de la Santé etde la Recherche Médicale U 97, Paris, France. Correspondence to Pierre-François Plouin, Assistance Publique-Hôpitaux de Paris, Hôpital Européen Georges Pompidou, rue Leblanc, 7598 Paris Cedex 15, France. pierre-francois.plouin@egp.aphp.fr American Heart Association, Inc. Hypertension is available at DOI:.1161/HYPERTENSIONAHA

2 526 Hypertension September the risk of complications. Our third objective was to explore variation in outcomes across subgroups. Methods Selection Criteria and Search Method for Identifying Studies Studies were considered eligible if: (1) they enrolled hypertensive patients with fibromuscular RAS; (2) patients underwent renal artery revascularization by surgical reconstruction or PTRA; and (3) the number of patients with hypertension cure could be identified. We searched Medline and Embase for all of the studies published before and including October 9. Complete search strategies are available in the online Data Supplement (please see Table S1). We screened the reference lists of primary articles and relevant reviews. We also screened textbooks on vascular medicine and interventional radiology, as well as conference proceedings from the Society of Interventional Radiology and proceedings from the Cardiovascular and Interventional Radiological Society of Europe annual meetings ( 8). Studies in languages other than English were not. Case reports and studies concerning 5 patients were. 781 records on the basis of title 143 records on the basis of abstract 91 records on the basis of full-text article 86 records identified through Medline and Embase database searching 35 abstracts screened 162 full-text articles assessed for eligibility 98 eligible articles 42 records identified through hand searching (reference lists, textbooks, conference proceedings) 15 records on the basis of full-text article Selection of Studies and Data Collection Two authors independently applied the selection criteria to all of the reports identified by the search strategy. Final selection was based on a review of full-text articles. Any disagreement regarding selection was discussed to obtain consensus. Only the most recent article was analyzed for multiple publications of a given study or when we could not exclude the possibility of overlapping populations. Two authors extracted data concerning baseline patient characteristics (age, sex, BP, and known duration of hypertension), FMD characteristics (criteria for FMD diagnosis, unilateral or bilateral disease, and medial or nonmedial FMD type), and revascularization techniques. Criteria for FMD diagnosis were present if renal angiography, the appearance of FMD lesions after angiography (medial-type FMD with multiple stenoses and the string-of-beads appearance or focal of tubular type FMD lesions), or histological confirmation was reported. Studies were judged to be at lower risk of bias in cases of prospective recruitment, consecutive recruitment, adequate description of the study population (at least mean age, mean baseline BP, and clinical criteria for revascularization), complete follow-up (hypertension cure assessed in % of patients with a successful procedure), and adequate outcome measurement (hypertension cure defined and assessed 1 days after revascularization for all of the patients). Outcome and Data Analysis The primary outcome was the rate of hypertension cure defined according to each study. Secondary outcomes included the technical success rate, the rate of periprocedural complications, the rate of hypertension cure or improvement (improvement being defined according to each study), and the effect of revascularization on renal function. Complications were classified according to localization (renal, puncture site, or cardiovascular complications) and severity (major or minor complications).,11 Renal complications included renal failure, embolization or infarction, and renal artery dissection, occlusion, thrombosis, perforation, or rupture. Puncture site complications included local hematoma or infection. Cardiovascular complications included nonrenal embolization or dissection. Major complications were those resulting in an additional procedure, unplanned treatment, prolonged hospitalization, transfusion, or death. For the meta-analysis, proportions were transformed into quantities via the Freeman-Tukey arcsine transformation. We computed a weighted mean of the individual transformed proportions using DerSimonian-Laird weights for the random effects model; the combined proportion was calculated through back transformation of the weighted mean. Inconsistencies in results across studies were assessed using the Cochran homogeneity 2 test and I 2 statistic. 19 because of possible overlapping populations 69 studies concerning PTA 5 studies concerning PTA 4 studies concerning surgery 25 studies concerning surgery 15 because of possible overlapping populations Figure 1. Flow diagram mapping out the numbers of records identified and included. To analyze variations in BP outcome, we first performed indirect comparisons between reports grouped according to the definition of hypertension cure. We then performed a sensitivity analysis to assess the small study effect on outcome rates. Because small-sized studies are prone to overestimating or underestimating BP outcome, we separated trials according to the median number of patients per study. We then performed subgroup analyses comparing patients with bilateral and unilateral RAS, those with medial and nonmedial FMD, and those with branch and nonbranch stenosis. Lastly, we performed meta-regression analyses to assess the influence of mean age, known duration of hypertension, baseline BP, and duration of follow-up on the outcome. Meta-regression models were used to assess the potential changes in rates of technical success, complications, and hypertension cure with time assessed based on the publication year. The meta-regression model was a logistic-normal model with binomial data and random effects. Results Selection of Studies Of the 86 records identified from our electronic search, 71 articles were deemed eligible. We identified an additional 27 eligible articles through manual searches. These 98 articles reported results from 69 series of patients treated with PTRA and from 4 series of patients treated with surgery. Among these, we 19 PTRA and 15 surgery studies over possible overlapping populations (please see Table S2). Finally, 5 studies concerning PTRA and 25 studies concerning surgery were included (Figure 1). Of these, 3 PTRA and 2 surgery studies specifically concerned pediatric populations

3 Trinquart et al Revascularization of FMD Renal Artery Stenosis 527 Table 1. Patient Characteristics in FMD Patients Treated by PTRA or Surgery PTRA Surgery Available Data Grand Mean Available Data Grand Mean No. of No. of (Range of Means No. of No. of (Range of Means Variable Studies Patients Across Studies) Studies Patients Across Studies) Age, y (28 to 59) (24 to 45) Men % ( to 5) % ( to 43) Bilateral RAS % ( to 67) % (6 to 48) Systolic/diastolic BP, mm Hg (145 to 192)/8 (89 to 143) (157 to 7)/112 (1 to 128) No. of antihypertensive drugs (1.2 to 3.3) (1.7 to 2.2) Known duration of hypertension, y (2.5 to 13) (3.3 to.) Medial FMD % (26 to 85) % (57 to 82) (please see Tables S3 and S4 for the list of references to selected studies and reasons for excluding full-text articles). Results of PTRA Study Characteristics Among the 47 studies relating to PTRA, 3 were prospective, 14 reported consecutive patients, 18 provided the minimum required information on the study population, and 38 reported a complete follow-up of patients. Hypertension cure was defined in 41 studies and was assessed 1 days after revascularization in 22 studies. In the 47 series consisting of 1616 patients with RAS caused by FMD and treated with PTRA, the median number of patients per series was (range: 5 to 442). Patient Characteristics FMD diagnosis criteria were reported in only 23 studies. Baseline characteristics were not reported exhaustively. Where reported, the mean age was 42 years, the proportion of men was 19%, the proportion of patients with bilateral RAS was 26%, and the proportion with medial FMD was 67% (Table 1). The minimal stenosis grade leading to PTRA was reported in 14 series: it was 25% in 1, 5% in 5 (and/or a 15% peak systolic pressure gradient in 1 series), 6% in 3, 7% in 2, 75% in 2, and was 8% in 1 study. We found no mention of poststenotic dilatation or renin analysis to define high-grade stenosis. Detailed characteristics of individual studies are available in the online Data Supplement (please see Table S5). Procedure Outcome The description of procedures was frequently missing. 5-F sheaths were used in 3 series, 6-F sheaths in 1 series, and both 5-F and 7-F sheaths were used in another series. Seven studies involved.35-in guide wires, and the procedure was performed through the femoral puncture site for all of the patients in 4 studies, for 99% of patients in 1 study (humeral route in 1 patient), and 8% of patients in another study (axillary artery access in % of patients). Stents were used in 3 studies and involved a total of 8 patients. Technical success rates were reported in 27 studies (838 patients; 997 procedures). The combined success rate was 88.2% [95% CI: 83.5 to 92.2], with considerable heterogeneity (range: 69.2% to.%; homogeneity test P.1; I 2 77% [95% CI: 67% to 84%]; see Figure S1). The meta-regression analysis showed no significant effects of time on the rate of technical success: the odds ratio (OR) associated with a publication date years later was 1.37 (95% CI:.74 to 2.54; P.32). Complications Complications were reported in studies (663 patients). The combined complication rate was 11.8% [95% CI: 8.2% to 15.9%] with substantial heterogeneity (range:.% to 42.9%; homogeneity test P.16; I 2 55% [95% CI: 16% to 72%]; Figure 2). The major and minor complication rates were 6.3% [95% CI: 4.1% to 9.%] and 5.3% [95% CI: 2.7% to 8.7%], respectively. The risk of mortality was.9% [95% CI:.3% to 1.7%]. The rates of puncture site and kidney complications were 3.4% [95% CI: 2.% to 5.1%] and 8.3% [95% CI: 5.% to 12.4%], respectively. Detailed forest plots for each complication are presented in the online Data Supplement (see Figure S2). The meta-regression analysis showed a nonsignificant trend toward an increased risk of complications with time (OR associated with a publication date years later: 1.63 [95% CI:.93 to 2.83]; P.). BP Outcome Hypertension cure rates were reported in the 47 series (1426 patients with follow-up). Follow-up durations showed large Complications (%) % (8.2 to 15.9) 3.4% (2. to 5.1) 8.3% (5. to 12.4) 1.2% (.5 to 2.3).9% (.3 to 1.7) 6.3% 5.3% (4.1 to 9.) (2.7 to 8.7) Puncture All Renal CV Death Major Minor Site Studies (663 patients) Figure 2. Meta-analysis of periprocedural complication rates after PTRA.

4 528 Hypertension September 8 All Studies Studies with or more patients Hypertension Cures (%) % (39.8 to 51.7) 35.8% (25.5 to 46.8) 49.% (4.6 to 57.6) 4% (31 to 48) 26.7% (17. to 37.7) 41% (3 to 52) Figure 3. Meta-analysis of hypertension cure rates after PTRA. 47 studies 1426 patients I 2 =77% (67 to 84) 11 studies 282 patients I 2 =69% (42 to 83) 24 studies 552 patients I 2 =74% (62 to 83) 22 studies 1142 patients I 2 =87% (82 to 91) 5 studies 5 patients I 2 =6% ( to 85) 11 studies 419 patients I 2 =81% (67 to 89) Definition of cure Any BP <14/9 mmhg without Rx DBP <9 mmhg without Rx Any BP <14/9 mmhg without Rx DBP <9 mmhg without Rx intrastudy and interstudy variation (mean follow-up was from 1 to 99 months). The definition for cure varied between studies: it was defined as BP levels 14/9 mm Hg without antihypertensive treatment in 11 studies; diastolic BP 9 mm Hg without treatment in 24 studies; diastolic BP 9 mm Hg (respectively, mm Hg) without treatment when patient age was 4 years (respectively, 4 years) in 1 study; a -mm Hg decrease in diastolic BP to a level 9 mm Hg in 2 studies; normotension without treatment in 5 studies; BP 15/9 mm Hg without treatment in 2 studies; and BP 16/95 mm Hg without treatment in 1 study. The definition was not clear in 1 study. Overall, the combined cure rate was 45.7% [95% CI: 39.8% to 51.7%], with considerable heterogeneity (range: 14% to %; homogeneity test P.1; I 2 78% [95% CI: 71% to 83%]; Figure 3). The definition of hypertension improvement also varied considerably across studies. The combined rate for cure or improvement was 86.4 [95% CI: 83.2 to 89.3], with substantial heterogeneity (range: 56% to %; homogeneity test P.1; I 2 58% [95% CI: 39% to 69%]). Detailed forest plots for each outcome are reported in the online Data Supplement (please see Figures S3 and S4). The rate of cure varied according to the definition of cure. It was 35.8% [95% CI: 25.5% to 46.8%] in the 11 studies using BP 14/9 mm Hg without treatment compared with 49.% [95% CI: 4.6% to 57.6%] in the 24 studies using diastolic BP 9 mm Hg without treatment (P.7). The sensitivity analysis restricted to the 22 studies with patients showed lower combined rates than the combined rates estimated for all 47 of the studies (Figure 3). In particular, the combined cure rate was 26.7% [95% CI: 17.% to 37.7%] in the 5 studies in which BP 14/9 mm Hg without treatment was used to define hypertension cure. Subgroup Analyses Combining the subgroup results from 6 trials (166 patients) showed no statistically significant difference in the cure rate between patients with bilateral and those with unilateral disease (combined risk ratio:.79 [95% CI:.36 to 1.77]; P.57), with substantial heterogeneity (homogeneity test P.3; I 2 6% [95% CI: % to 82%]; please see Figure S5). Combining the results from 2 trials (77 patients) showed a lower probability of cure in cases of medial FMD than in nonmedial FMD (combined risk ratio:.6 [95% CI:.39 to.93]; P.2), with moderate heterogeneity (homogeneity test P.32; I 2 not applicable; please see Figure S6). Combining the results from 2 trials (129 patients) showed no statistically significant difference in the cure rate between patients with branch stenosis and those with other location stenoses (combined risk ratio:.89 [95% CI:.38 to 2.6]; P.78), without heterogeneity (homogeneity test P.83; I 2 not applicable; please see Figure S7). Finally, combining the results from 2 trials (89 patients) showed that the mean age was significantly lower among cured patients (combined standardized mean difference of.99 [95% CI: 1.54 to.43]; corresponding with a difference in mean age of years; P.5) without heterogeneity (homogeneity test P.98; I 2 not applicable). Similarly, the mean known duration of hypertension in these 2 studies tended to be lower among cured patients (combined standardized mean difference of.48 [95% CI: 1.2 to.6]; P.8) and without heterogeneity (homogeneity test P.63; I 2 not applicable). Meta-Regression Analyses Meta-regression analyses are reported in Table 2 and Figure 4. The probability of cure significantly decreased with increasing mean age (OR associated with an increase in mean age of years:.48 [95% CI:.39 to.59]) and mean known duration of hypertension (OR associated with an increase in mean hypertension duration of 5 years:.39 [95% CI:.23 to.67]). Finally, the probability of cure significantly decreased with time, as evaluated via the publication

5 Trinquart et al Revascularization of FMD Renal Artery Stenosis 529 Table 2. Univariate Meta-Regression Analyses Assessing the Relationship Between Hypertension Cure Rate and Patient or Study Characteristics in Series of FMD Patients Treated by PTRA or Surgery Variable Increase of Odds Ratio (95% CI) P PTRA Mean age, y.48 (.39 to.59).1 Mean systolic BP, mm Hg (.35 to 1.59).46 Mean diastolic BP, mm Hg (.95 to 3.47).9 Mean No. of antihypertensive drugs 1.57 (.24 to 1.39).24 Mean duration of hypertension, y 5.39 (.23 to.67).5 Mean follow-up, mo 6 1. (.99 to 1.1).97 Publication year.62 (.45 to.85).4 Surgery Mean age, y.84 (.58 to 1.23).39 Mean follow-up, mo 6.97 (.91 to 1.3).36 Publication year.81 (.62 to 1.7).15 year (OR associated with an increase of years:.62 [95% CI:.45 to.85]). Repeat PTRA Repeated PTRA between initial revascularization and hypertension cure was reported in series (357 patients who underwent follow-up). In these series, the combined proportion of patients receiving a secondary procedure was estimated to be 18.2% [95% CI: 11.% to 26.8%], with considerable heterogeneity (range: 8% to 56%; homogeneity test P.1; I 2 68% [95% CI: 23% to 82%]). Renal Outcome The effect of revascularization on renal function was assessed in 7 studies (7 patients) and showed no differences between mean creatinine levels at baseline and at follow-up (please see Table S6). Pediatric Cases Three series specifically involved pediatric populations with FMD treated using PTRA (47 patients with a mean age of 9 years, of which 49% were men). In 1 study, hypertension cure was defined by a BP lower than that of the 95th percentile for age and sex without treatment and assessed at a minimum of 6 months after PTRA; all 3 children (%) were completely cured of hypertension, with no complications, and remained hypertension free for 2 years after PTRA. In the other 2 series, hypertension cure was defined as normotensive without treatment and was assessed after a mean follow-up of 14 months (range: 1 to 6 months) and 87 months (range: 3 to 129 months), respectively. The hypertension cure rates were 67% (8 of 12 children) and 8% (4 of 5), respectively. Results of Surgery Study Characteristics Among the 23 series, there were 2 prospective studies, 3 with consecutive patients, 3 with the minimum required information on the study population, with a complete follow-up, in which hypertension cure was defined, and 15 in which Probability of hypertension cure (%) Probability of hypertension cure (%) Mean age (years) Publication year hypertension cure was measured 1 days after revascularization. In the 23 series consisting of 14 patients with RAS caused by FMD and treated using surgery, the median number of patients per series was 33 (range: 6 to 179). Several techniques were used in each series, the most frequent being aortorenal bypass graft using a saphenous vein or a prosthesis. Aortic reimplantation, resect anastomosis, and autotransplantation were also used. Eleven series included patients treated by primary nephrectomy. Separate analyses of BP outcome for each technique were not possible. Patient Characteristics FMD diagnosis criteria were reported in 14 studies. Where provided, the mean age was 36 years, and the proportion of men was %. The proportion of patients with bilateral RAS was 23%, and the proportion of patients with medial FMD was 7% (Table 1). Detailed characteristics of individual studies are available in the online Data Supplement (please see Table S7). Complications Mortality data were reported in 18 studies (778 patients), of which 16 series reported no death, yielding a combined Figure 4. Meta-regression analyses assessing the relationship between the hypertension cure rate after PTRA and mean age, publication year

6 53 Hypertension September perioperative risk of death estimated to be 1.2% [95% CI:.5% to 2.1%]. Complication rates were reported in 5 studies (222 patients). The combined complication rate was 16.9% [95% CI:.3% to 24.7%], with substantial heterogeneity (range: 5% to 28%; homogeneity test P.8; I 2 52% [95% CI: % to 81%]). The majority of reported complications were classified as major complications, and the major complication rate was estimated to be 15.4% [95% CI: 9.2% to 22.8%]. Detailed forest plots for each complication can be found in the online Data Supplement (see Figure S8). BP Outcome Assessing revascularization success by follow-up angiography was not mentioned. The hypertension cure rate was reported in the 23 selected series (5 patients with followup). Follow-up durations showed large variation: 4 series reported the immediate postoperative outcome, and in the other series, the mean follow-up ranged from 8 to months. The definition of cure varied across studies: it was defined as BP 14/9 mm Hg without treatment in 5 studies; diastolic BP 9 mm Hg without treatment in 12 studies; a -mm Hg decrease in diastolic BP to a level 9 mm Hg in 1 study; normotension without treatment in 2 studies; BP 15/9 mm Hg without treatment in 1 study; and diastolic BP mm Hg without treatment in 1 study. The definition was not clear in 1 study. Overall, the combined cure rate was 57.5% [95% CI: 53.% to 62.%], with substantial heterogeneity (range: 14% to %; homogeneity test P.1; I 2 47% [95% CI: 13% to 67%]). The definition of hypertension improvement also varied considerably across studies. The combined rate of cure or improvement was 88.3% [95% CI: 83.2% to 92.6%], with considerable heterogeneity (range: 62% to %; homogeneity test P.1; I 2 8% [95% CI: 7% to 86%]). Detailed forest plots for each outcome are reported in the online Data Supplement (please see Figures S9 through S11). The rate of hypertension cure tended to be lower in the 5 series (29 patients) that used BP 14/9 mm Hg without treatment than in the 12 series (442 patients) that defined cure as diastolic BP 9 mm Hg without treatment: 54.1% (95% CI: 44.1% to 64.%) versus 59.8% (95% CI: 52.8% to 66.7%; P.35). The sensitivity analysis restricted to the 12 studies with 33 patients (824 patients) showed a similar combined cure rate to that estimated using all 23 studies: 56.6% (95% CI: 5.3% to 62.8%). Moreover, the sensitivity analysis restricted to the studies without patients treated by nephrectomy (47 patients) also showed a similar combined rate (57.5% [95% CI: 51.1% to 63.7%]). Subgroup analyses were reported in a single series, which found no differences in cure rates between patients with bilateral and unilateral disease (15 of 27 versus 16 of 29; risk ratio: 1. [95% CI:.63 to 1.61]; P 1.) or cure rates between patients with medial and nonmedial lesions (18 of 32 versus 13 of 24; risk ratio: 1.4 [95% CI:.64 to 1.68]; P 1.). Meta-Regression Analysis Meta-regression analyses showed that the probability of hypertension cure tended to decrease with increasing mean age (OR associated with an increase in mean age of years:.84 [95% CI:.58 to 1.23]) and time, as evaluated via the publication year (OR associated with an increase of years:.81 [95% CI:.62 to 1.7]), but that these changes were not statistically significant (Table 2). Pediatric Cases Two series specifically concerned pediatric populations (6 patients; mean age: 12 years; men: 53%). Most children underwent aortorenal bypasses using saphenous vein (62%), whereas some underwent a nephrectomy (22%). Two retroperitoneal hematomas and a fatal thrombosis were reported in 1 study involving 27 children, and no complications were reported in the other. In 1 study, hypertension cure was defined as BP 15/9 mm Hg without treatment after a minimum follow-up of 6 months; in the other, it was defined as BP 14/9 mm Hg without treatment after a minimum follow-up of 12 months. Hypertension cure rates were 59.2% (16 of 27 children) and 87.9% (29 of 33), respectively. Discussion This systematic review identified 7 series of patients with RAS caused by FMD who underwent revascularization using PTRA or surgery (47 and 23 series totaling 1616 and 14 patients, respectively) and in whom postintervention follow-up allowed the assessment of BP outcome and, in most series, how often complications occurred. Most patients were women in their 4s. One patient in 4 had bilateral FMD. Hypertension cure after PTRA or surgery was obtained in 46% and 55% of patients, respectively, with large variations in cure rates across studies. The probability of cure was negatively associated with age, known duration of hypertension, medial-type FMD, time of publication, and more stringent definitions of cure. Cure rates based on current definition (BP 14/9 mm Hg without treatment) were only 36% and 54%, respectively, after PTRA and surgery. The risk of periprocedural complications was substantial, estimated at 12% after PTRA and 17% after surgery. However, major complications tended to be less frequent after PTRA than after surgery (6% versus 15%), but selection criteria may have differed between study populations. The risk of death was estimated to be.9% after PTRA and 1.2% after surgery. The earlier review of PTRA published by Ramsay and Waller ( series, 193 patients with FMD RAS) 4 reported rates of cure of 5%. A more recent review by Matsumoto et al (26 series, 552 patients) 12 reported a mean rate of cure of 42% for PTRA, and a review by Schneider et al ( series, 759 patients undergoing surgery) 13 reported a mean rate of cure of 54% for surgery. Our current systematic review suggests a less favorable benefit:risk ratio for revascularization in patients with RAS attributed to FMD than previous nonsystematic reviews. The limited benefit of revascularization in relatively young subjects with normal renal function in virtually all cases has several plausible explanations. First, screening for FMD has considerably improved in the last decade because of the widespread use of computed tomography or magnetic resonance angiography when screening for secondary hypertension. The string-of-beads appearance, which can be detected noninvasively by these techniques, is a specific indicator of FMD. 14 However, the quantification of FMD lesions using

7 Trinquart et al Revascularization of FMD Renal Artery Stenosis 531 noninvasive techniques is difficult. 15 By contrast, the prevalence of FMD lesions in normotensive subjects, such as potential kidney donors, is relatively high It may be high as well in patients with essential hypertension. Consequently, patients with renal FMD lesions and essential hypertension may be advised to undergo PTRA, resulting in a disappointing BP outcome. This is a plausible explanation for the negative trend in the cure rate and the date of publication. Second, renal revascularization may be incomplete because of restenosis. A substantial proportion of patients received a repeat PTRA in the present series. The incidence of restenosis was also analyzed in cohort studies specifically designed to assess renal artery patency after PTRA in FMD. One study reported that restenosis levels were 6% at 12 months in 7 (25%) of 27 patients, and another study reported that 4 (11%) of 34 patients had displayed restenosis after a 3- to 25-month follow-up that was 1 grade or higher than at the end of the procedure. 21 Third, hypertension persistence could be explained by bilateral FMD lesions or the presence of a systemic disease. Nearly 25% of the patients analyzed in the present series had bilateral disease, and subgroup analysis showed a trend toward a reduced cure rate in the bilateral group. FMD is a heterogeneous disease, and its classification into subtypes is based on pathological or angiographic examination Because most patients currently undergo PTRA rather than surgery, pathological specimens are rarely available. The angiographic classification distinguishes the multifocal string-of-beads subtype (mostly medial) from unifocal or tubular subtypes (mostly nonmedial). This classification was reported in only 7 of 47 series relating to PTRA and in only 3 of 23 series relating to surgery. When specified, 7% of patients treated using PTRA or surgery had medial FMD disease. BP outcome was less favorable in patients with medial-type FMD than in those with unifocal or tubular FMD. Patient age and bilateral disease are confounding variables, because mean age is much lower and bilateral disease is less frequent in nonmedial than in medial FMD. 25 In series concerning children, the BP outcome tended to be good cure rates of 67% to % after PTRA and 6% to 88% after surgery, although the evidence is based on small numbers of patients. Our systematic review has strengths and limitations. We used a rigorous, replicable, and transparent method and identified a significantly greater number of patient series than previous reviews. 4,12,13 No languages were restricted, and we searched through gray literature. Using funnel plots, we found no evidence for publication bias (please see Figure S12). There are several limitations within any meta-analysis. Combined estimates should be interpreted cautiously because of the considerable heterogeneity across studies. Multiple definitions of hypertension cure were used, times from revascularization to follow-up varied within and across studies, and there was diversity in patient case mix, reflecting possible differences in selection criteria. Surgical series involved multiple techniques, including nephrectomy. We incorporated heterogeneity into the random effects models, and the 95% CI describes the uncertainty in the mean probability of hypertension cure. We performed sensitivity analyses. For PTRA, it revealed that small studies tended to overestimate the cure rate. Therefore, more confidence can be drawn from estimates issued from studies with patients. For surgery, similar estimates were obtained when the metaanalysis was restricted to studies including more than the median number of patients and to studies excluding patients with nephrectomy as the primary treatment. The analysis of sources of heterogeneity (subgroup and meta-regression analyses) may experience confounding variables, which is a significant problem in meta-analysis of observational studies, because these analyses are univariate. Finally, data on complications may have experienced selective reporting, and we cannot exclude publication bias, which could result in overestimating the hypertension cure rate. Nevertheless, the identified factors associated with hypertension cure are plausible: increasing age and duration of hypertension are associated with an increasing chance of renal parenchymal disease, concurrent atherosclerotic lesions or alteration in aortic compliance, and a lower chance of cure. The trend toward a decreasing probability of cure after revascularization by PTRA or surgery over time can be explained by easier screening for FMD, leading to identification of less severe forms of the disease and to widening of the indications for revascularization. Perspectives Further prospective studies are required. Well-designed cohort studies should address the identification of prognostic factors for hypertension cure, such as FMD type or bilateral disease. Standardizing the definition of BP cure (using the recognized criteria of BP inferior to 14/9 mm Hg without treatment) and using outcome assessment at a fixed follow-up for all patients are mandatory. Considering that only 1 patient in 3 has normal BP after PTRA with a 12% risk of complication, a randomized trial comparing PTRA with medical treatment for FMD should be considered. Acknowledgments We thank Sophie Guiquerro for literature search support. None. Disclosures References 1. Slovut DP, Olin JW. Fibromuscular dysplasia. N Engl J Med. 4;35: Dworkin LD, Cooper CJ. Clinical practice: renal-artery stenosis. N Engl J Med. 9;361: Streeten DH, Anderson GH Jr, Wagner S. Effect of age on response of secondary hypertension to specific treatment. Am J Hypertens. 199;3: Ramsay LE, Waller PC. Blood pressure response to percutaneous transluminal angioplasty for renovascular hypertension: an overview of published series. BMJ. 199;3: Uzzo RG, Novick AC, Goormastic M, Mascha E, Pohl M. Medical versus surgical management of atherosclerotic renal artery stenosis. Transplant Proc. 2;34: Ives NJ, Wheatley K, Stowe RL, Krijnen P, Plouin PF, van Jaarsveld BC, Gray R. Continuing uncertainty about the value of percutaneous revascularization in atherosclerotic renovascular disease: a meta-analysis of randomized trials. Nephrol Dial Transplant. 3;18: Nordmann AJ, Logan AG. Balloon angioplasty versus medical therapy for hypertensive patients with renal artery obstruction. Cochrane Database Syst Rev. 3;(3):CD2944.

8 532 Hypertension September 8. Balk E, Raman G, Chung M, Ip S, Tatsioni A, Alonso A, Chew P, Gilbert SJ, Lau J. Effectiveness of management strategies for renal artery stenosis: a systematic review. Ann Intern Med. 6;145: Wheatley K, Ives N, Gray R, Kalra PA, Moss JG, Baigent C, Carr S, Chalmers N, Eadington D, Hamilton G, Lipkin G, Nicholson A, Scoble J. Revascularization versus medical therapy for renal-artery stenosis. N Engl J Med. 9;361: Rundback JH, Sacks D, Kent KC, Cooper C, Jones D, Murphy T, Rosenfield K, White C, Bettmann M, Cortell S, Puschett J, Clair DG, Cole P. Guidelines for the reporting of renal artery revascularization in clinical trials. J Vasc Interv Radiol. 3;14:S477 S Leoni CJ, Potter JE, Rosen MP, Brophy DP, Lang EV. Classifying complications of interventional procedures: a survey of practicing radiologists. J Vasc Interv Radiol. 1;12: Matsumoto AH, Spinosa DJ, Angle F, Hagspiel KD, Leung DA. Evaluation and endovascular therapy for renal artery stenosis. In: Baum S, Pentecost MJ, eds. Abram s Angiography: Interventional Radiology. 2nd ed. Philadelphia, PA: Lippincott Williams & Wilkins; 6: Schneider DB, Stanley JC, Messina MM. Renal artery fibrodysplasia and renovascular hypertension. In: Rutherford RB, ed. Vascular Surgery. 5th ed. Philadelphia, PA: W.B. Saunders; : Vasbinder GB, Nelemans PJ, Kessels AG, Kroon AA, Maki JH, Leiner T, Beek FJ, Korst MB, Flobbe K, de Haan MW, van Zwam WH, Postma CT, Hunink MG, de Leeuw PW, van Engelshoven JM. Accuracy of computed tomographic angiography and magnetic resonance angiography for diagnosing renal artery stenosis. Ann Intern Med. 4;141: ; discussion Willoteaux S, Faivre-Pierret M, Moranne O, Lions C, Bruzzi J, Finot M, Gaxotte V, Mounier-Vehier C, Beregi JP. Fibromuscular dysplasia of the main renal arteries: comparison of contrast-enhanced MR angiography with digital subtraction angiography. Radiology. 6;241: Spring DB, Salvatierra O Jr, Palubinskas AJ, Amend WJ Jr, Vincenti FG, Feduska NJ. Results and significance of angiography in potential kidney donors. Radiology. 1979;133: Cragg AH, Smith TP, Thompson BH, Maroney TP, Stanson AW, Shaw GT, Hunter DW, Cochran ST. Incidental fibromuscular dysplasia in potential renal donors: long-term clinical follow-up. Radiology. 1989; 172: Neymark E, LaBerge JM, Hirose R, Melzer JS, Kerlan RK Jr, Wilson MW, Gordon RL. Arteriographic detection of renovascular disease in potential renal donors: incidence and effect on donor surgery. Radiology. ;214: Andreoni KA, Weeks SM, Gerber DA, Fair JH, Mauro MA, McCoy L, Scott L, Johnson MW. Incidence of donor renal fibromuscular dysplasia: does it justify routine angiography? Transplantation. 2;73: Birrer M, Do DD, Mahler F, Triller J, Baumgartner I. Treatment of renal artery fibromuscular dysplasia with balloon angioplasty: a prospective follow-up study. Eur J Vasc Endovasc Surg. 2;23: Plouin PF, Darne B, Chatellier G, Pannier I, Battaglia C, Raynaud A, Azizi M. Restenosis after a first percutaneous transluminal renal angioplasty. Hypertension. 1993;21: Harrison EG Jr, McCormack LJ. Pathologic classification of renal arterial disease in renovascular hypertension. Mayo Clin Proc. 1971;46: Kincaid OW, Davis GD, Hallermann FJ, Hunt JC. Fibromuscular dysplasia of the renal arteries: arteriographic features, classification, and observations on natural history of the disease. Am J Roentgenol Radium Ther Nucl Med. 1968;4: McCormack LJ, Poutasse EF, Meaney TF, Noto TJ Jr, Dustan HP. A pathologic-arteriographic correlation of renal arterial disease. Am Heart J. 1966;72: Pannier-Moreau I, Grimbert P, Fiquet-Kempf B, Vuagnat A, Jeunemaitre X, Corvol P, Plouin PF. Possible familial origin of multifocal renal artery fibromuscular dysplasia. J Hypertens. 1997;15: