Accepte

Reprintsity of Pdelphia,

0039-60

� 2014

http://d

1394

Long-term blood pressure control inpatients undergoing adrenalectomyfor primary hyperaldosteronismHeather Wachtel, MD,a Isadora Cerullo, BA,a Edmund K. Bartlett, MD,a Rachel R. Kelz, MD,a

DebbieL.Cohen,MD,bGiorgosC.Karakousis,MD,aRobertE.Roses,MD,a andDouglasL. Fraker,MD,a

Philadelphia, PA

Background. Data on long-term blood pressure (BP) control after adrenalectomy for primaryhyperaldosteronism are limited. We analyzed long-term outcomes to identify factors predictive of cure.Methods. We performed a retrospective cohort study of patients undergoing adrenalectomy for primaryhyperaldosteronism (1997–2013). BP and antihypertensive medications were assessed at long-termfollow-up ($12 months). Primary outcome was cure, defined as normotension off antihypertensives.Results. Of 85 patients, 15.3% (n = 13) were cured, 54.1% (n = 46) were normotensive while re-maining on anti-hypertensives, and 30.6% (n = 26) were hypertensive. Younger age (P = .011), femalesex (P < .001), lesser body mass index (P = .018), shorter duration of hypertension (P = .002), lowercreatinine (P = .001), and fewer preoperative antihypertensive medications (P < .001) were associatedwith cure. Female sex, body mass index #25 kg/m2, hypertension <5 years, creatinine #0.8 mg/dL,and <2 antihypertensives were incorporated into a scoring system. For a score of 0–1 (n = 61) the curerate was 3%; 100% of patients with a score of 4–5 (n = 3) were cured. This scoring system performedcomparably to the Aldosterone Resolution Score, which has been used to evaluate short-term postoperativeoutcomes.Conclusion. This is the largest study to identify factors associated with long-term BP control afteradrenalectomy and incorporate these into a scoring system. These data provide a potential tool to guidepreoperative patient counseling. (Surgery 2014;156:1394-403.)

From the Department of Surgerya and Division of Renal, Electrolyte, and Hypertension,b Hospital of theUniversity of Pennsylvania, Philadelphia, PA

PRIMARY HYPERALDOSTERONISM is one of the few poten-tially curable causes of hypertension. Previouslythought to be rare, it has now been identified inapproximately 11–20% of patients with hyperten-sion resistant to medical management.1,2 Thelong-term sequelae of untreated hypertensionhave been documented extensively and includepotentially irreversible cardiovascular and renaldamage. Multiple studies suggest that target organdamage is more extensive,3-5 and cardiovascularevents more common,6,7 in the setting of hyperten-sion secondary to primary hyperaldosteronism.Indeed, primary hyperaldosteronism has beenshown to cause target organ dysfunction indepen-dent of hypertension.8 Early detection and

d for publication August 11, 2014.

requests: Heather Wachtel, MD, Hospital of the Univer-ennsylvania, 3400 Spruce Street, 4 Maloney, DSE, Phila-PA 19104. E-mail: heather.wachtel@uphs.upenn.edu.

60/$ - see front matter

Elsevier Inc. All rights reserved.

x.doi.org/10.1016/j.surg.2014.08.021

SURGERY

appropriate treatment of primary hyperaldosteron-ism is therefore of paramount importance in pre-venting long-term morbidity and mortality.

Idiopathic bilateral adrenal hyperplasia is themost common cause of primary hyperaldosteron-ism; aldosterone-producing adenoma is the secondmost frequent, accounting for approximately 35%of cases.9 Other less-common causes include uni-lateral adrenal hyperplasia, aldosterone-secretingadrenocortical carcinoma, glucocorticoid-remedi-able hyperaldosteronism, and ectopic aldosteronesecreting tumors. Whereas bilateral hyperplasia isbest managed medically, ipsilateral adrenalectomyis the standard of care for unilateral oversecretionof aldosterone caused by aldosterone-producingadenoma and unilateral hyperplasia.

Despite the potentially curable nature of pri-mary hyperaldosteronism caused by aldosterone-producing adenoma and unilateral hyperplasia,only approximately one-third of patients experi-ence complete normalization of blood pressure offall antihypertensive medications after surgery10,11;the remainder benefit from operative intervention,

SurgeryVolume 156, Number 6

Wachtel et al 1395

experiencing improved blood pressure control orrequiring fewer antihypertensive medications. Res-olution of hypertension typically is evaluated 6–12 months after surgery, and numerous studieshave identified factors associated with short-termoutcomes after adrenalectomy for primary hyperal-dosteronism11-13; however, data on long-termblood pressure control is limited. In this study,we evaluated long-term outcomes in our large se-ries of patients undergoing adrenalectomy for pri-mary hyperaldosteronism.

METHODS

We performed a retrospective cohort study of allpatients undergoing adrenalectomy for primaryhyperaldosteronism due to aldosterone-producingadenoma or unilateral hyperplasia at a tertiarymedical center during a 16-year period (1997–2013). Consecutive patients were identified in ourprospectively maintained endocrine surgery regis-try. Children (<18 years of age) were excluded.

Data abstracted included patient demographics,biochemical testing, antihypertensive medications,and blood pressure measurements. Patients werereferred routinely for adrenal vein sampling. Co-morbid renal disease was defined as chronic renalinsufficiency, chronic renal failure, or any knownrenal disease, which in this cohort included renalartery stenosis, fibromuscular dysplasia, nephroli-thiasis, and polycystic kidney disease.

Standard blood pressure criteria were used, withhypertension defined as $140/90 mm Hg. Post-operative follow-up data were obtained at thepatient’s clinic visit 2–3 weeks after operation.Short-term follow-up was defined as 6–12 monthsafter surgery. Long-term follow-up was definedas $12 months after surgery. Of 164 patients whounderwent surgery, 68.3% (n = 112) did not havefollow-up data in the medical record and werecontacted by telephone to obtain current bloodpressure measurements and antihypertensivemedication regimens. Patients who were unableto be reached or who did not have recent or exactblood pressure measurements were excluded fromthe study. Long-term follow-up data were obtainedfor a total of 85 patients (52%) who constitutedthe study population. Of the study population,61% (n = 52) had data in the medical record,whereas 39% (n = 33) had data obtained bytelephone follow-up. Of the study population, 31patients (37%) also had complete short-termfollow-up data available in the medical record.

Primary outcome was cure, defined as normo-tension off antihypertensive medications. Groupcomparisons were made using the Student t test, v2

test, or Wilcoxon rank sum test, as appropriate. Todichotomize continuous variables, receiver oper-ating characteristic (ROC) curves were generated,and values with maximal sensitivity were identified.For ease of use, values were rounded to the nearestinteger or clinically appropriate value. The 5 fac-tors associated with cure were incorporated intoan unweighted scoring system in which 1 pointwas assigned per risk factor, for a possible scorerange of 0–5 points.

The Aldosternoma Resolution Score (ARS) asdescribed by Zarnegar et al,13 is a 4-item, weightedscoring system (possible score range of 0–5points), which was developed and validated on pri-mary hyperaldosteronism patients 6 months afteradrenalectomy. To determine whether the ARSperformed well in long-term follow-up, we vali-dated it in our cohort in accordance with pub-lished data, and compared it to our own scoringsystem using ROC curves and 95% confidence in-tervals (CIs).

Institutional review board approval for thisstudy was obtained from the University of Pennsyl-vania. Statistical analysis was performed usingSTATA/IC 12.1 software (Stata Corporation, Col-lege Station, TX).

RESULTS

Cohort characteristics. Of 164 patients identi-fied for inclusion, long-term follow-up data wereobtained for 85 patients (51.8%), who wereincluded in the final analysis. Mean age was51.5 ± 10.9 years; 37.8% of patients were female(n = 62; Table I). The majority of patients werewhite (65.9%, n = 108), with black as the secondmost common ethnicity (23.2%, n = 38). The me-dian duration of preoperative hypertension was10 years (interquartile range 4–20 years). Onpreoperative evaluation, patients were on a meanof 3.5 ± 1.5 antihypertensive medications, witha mean systolic blood pressure (SBP) of152.5 ± 21.1 mm Hg. On postoperative follow-up2–3 weeks after surgery, patients were on adecreased number of antihypertensive medica-tions (1.8 ± 1.5) with an improved a mean SBP of136.2 ± 16.8 mm Hg. At long-term follow-up (me-dian 36 months, interquartile range 15–80 months)the mean number of antihyptertensive agents re-mained stable at 1.9 ± 1.5, whereas blood pressurecontrol had improved (mean SBP 129.9 ± 14.9 mmHg) although this was not statistically significant(P = .662). From preoperative to long-termfollow-up, the mean reduction in antihypertensiveagents was 1.8 ± 1.4, and mean reduction in SBPwas 23.2 ± 23.7 mm Hg.

Table I. Characteristics of patients undergoingadrenalectomy for primary hyperaldosteronism

Age, yMean (SD) 51.5 (10.9)

Sex, n (%)Female 62 (37.8)Male 102 (62.2)

BMI, kg/m2

Median (IQR) 31.7 (26.9–36.7)Ethnicity, n (%)

White 108 (65.9)Black 38 (23.2)Asian 8 (4.9)Hispanic 6 (3.7)Other 4 (2.4)

Duration of hypertension, yMedian (IQR) 10 (4–20)

Comorbid renal disease (%) 19 (11.7)Smoking history, n (%) 43 (26.2)Serum creatinine, mg/dL

Median (IQR) 1.0 (0.8–1.2)Serum aldosterone, ng/dL

Median (IQR) 31.1 (20.4–47)Serum renin activity, ng/mL/h

Median (IQR) 0.2 (0.1–0.4)Aldosterone:renin ratio, (ng/dL)/(ng/mL/h)

Mean (SD) 248.6 (533.5)Adrenal vein sampling

performed, n (%)74 (87.1)

Operative approach, n (%)Laparoscopic 80 (94.1)Open 2 (2.4)Laparoscopic converted to open 3 (3.5)

Tumor size, cmMedian (IQR) 1.5 (1.0–2.0)

Duration of follow-up, monthsMean (SD) 49.7 (44.0)Median (IQR) 36 (15–80)

Mean SBP, mm HgPreoperative (SD) 152.5 (21.1)Postoperative follow-up (SD) 136.2 (16.8)Short-term follow-up (SD)* 128.4 (11.2)Long-term follow-up (SD) 129.9 (14.9)Reduction at long-termfollow-up (SD)

23.2 (23.7)

Mean no. antihypertensivesPreoperative (SD) 3.5 (1.5)Postoperative follow-up (SD) 1.8 (1.5)Short-term follow-up (SD)* 2.4 (1.4)Long-term follow-up (SD) 1.9 (1.5)Reduction at long-termfollow-up (SD)

1.8 (1.4)

*Data available for 31 patients.BMI, Bodymass index; IQR, interquartile range;SBP, systolic bloodpressure.

SurgeryDecember 2014

1396 Wachtel et al

Blood pressure control, by cure status. Patientswere classified as cured (normotensive off allantihypertensive medication) or persistently hyper-tensive (patients who continued to require

antihypertensive medications). At long-termfollow-up, 15.3% (n = 13) of patients were curedwhereas 84.7% (n = 72) were persistently hyperten-sive (Table II). In the persistently hypertensivegroup, 63.9% (n = 46) were normotensive whilecontinuing to require antihypertensive medica-tions, and 36.1% (n = 26) were hypertensivedespite remaining on medications. Of patientswho were persistently hypertensive, 97.2%(n = 70) benefited from surgery and demonstratedeither a decrease in number of antihypertensivemedications, improved blood pressure control, orboth, after operation.

Mean SBP on long-term follow-up was119.9 ± 8.9 mm Hg for the cured group, versus137.7 ± 16.2 mm Hg for the group that waspersistently hypertensive (P = .012). In compari-son, on short-term follow-up, there was no differ-ence between cured and persistently hypertensivepatients in terms of mean SBP (129.8 ± 6.7 vs128.3 ± 11.8 mm Hg). The 2 groups had equivalentblood pressure reduction on long-term follow-upcompared with preoperative blood pressure(23.9 ± 20.0 mm Hg for cured patients vs23.7 ± 24.8 mm Hg for persistently hypertensive,P = .980).

In examining numbers of antihypertensiveagents, we found that patients in the cured cohortwere on a smaller number of medications (0.5 vs2.6, P = .002) at short-term follow-up comparedwith persistently hypertensive patients. This differ-ence continued to be relevant at long-term follow-up, when cured patients were by definition off allmedications, whereas persistently hypertensivepatients were on a mean of 2.4 ± 1.4 antihyperten-sives. The overall reduction in number of antihy-pertensive agents was comparable between curedand persistently hypertensive patients (2.3 ± 0.8vs 1.7 ± 1.5 medications, P = .097).

At short-term follow-up, cure rate was only 6.5%(n = 2). In analyzing the group (n = 31) with bothshort- and long-term follow-up data available, cureat short-term follow-up was highly specific (100%)but only 50% sensitive for cure at long-term follow-up. Positive predictive value was 100% and nega-tive predictive value was 93.1%. Two patientschanged status from not cured at the short-termtime point, to cured at the long-term time point.

Factors associated with cure on long-termfollow-up. On univariate analysis, younger age(P = .011), female sex (P<.001), lower bodymass in-dex (BMI; P = .018), shorter duration of hyperten-sion (P = .002), lesser serum creatinine (P = .001),and fewer preoperative antihypertensive medica-tions (P<.001) were associated with cure (Table II).

Table II. Factors associated with cure at long-term follow-up after adrenalectomy for primaryhyperaldosteronism

Cure (n = 13) Persistent hypertension (n = 72) P value

Age, yMean (SD) 45.7 (9.6) 54.0 (10.8) .011

Sex, n (%) <.001Female 11 (84.6) 20 (27.8)Male 2 (15.4) 52 (72.2)

BMI, kg/m2

Median (IQR) 28.0 (24.9–31.1) 33.7 (27.6–37.8) .018Ethnicity, n (%) .353

White 10 (76.9) 44 (61.1)Black 2 (15.4) 25 (34.7)Asian 1 (17.7) 2 (2.3)Hispanic 0 (0) 1 (1.4)Other 0 (0) 0 (0)

Duration of hypertension, yMedian (IQR) 4.0 (2.0–10) 12 (8–20) .002

Comorbid renal disease, n (%) 1 (7.7) 13 (18.1) .354Smoking history, n (%) 4 (33.3) 22 (32.4) .947Serum creatinine, mg/dL .001

Median (IQR) 0.8 (0.7–0.9) 1.1 (0.9–1.4)Serum aldosterone, ng/dL

Median (IQR) 34.0 (23.5–36.0) 31.4 (20.0–53.2) .926Serum renin activity, ng/mL/h

Median (IQR) 0.3 (0.2–0.5) 0.2 (0.1–0.3) .255Aldosterone:renin ratio, (ng/dL)/(ng/ml/h)

Mean (SD) 160.3 (118.2) 326.6 (769.5) .460Adrenal vein sampling performed, n (%) 10 (76.9) 64 (88.9) .237Operative approach, n (%) .300

Laparoscopic 12 (92.3) 68 (94.4)Open 1 (7.7) 1 (1.4)Laparoscopic converted to open 0 (0.0) 3 (4.2)

Tumor size, cmMedian (IQR) 1.5 (1.3–1.8) 1.4 (1.0–2.0) .503

Duration of follow-up, monthsMedian (IQR) 32.0 (15.0–107.0) 47.0 (22.0–86.0) .498

Mean SBP, mm HgPreoperative (SD) 143.8 (18.6) 154.7 (20.3) .078Postoperative follow-up (SD) 119.5 (2.1) 137.7 (16.2) .137Short-term follow-up (SD)* 129.8 (6.7) 128.3 (11.8) .182Long-term follow-up (SD) 119.9 (8.9) 131.7 (16.1) .012Reduction at long-term follow-up (SD) 23.9 (20.0) 23.7 (24.8) .980

Mean no. antihypertensivesPreoperative (SD) 2.3 (0.8) 4.0 (1.6) <.001Postoperative follow-up (SD) 0.5 (0.7) 2.3 (1.6) <.001Short-term follow-up (SD)* 0.5 (0.6) 2.6 (1.2) .002Long-term follow-up (SD) 0 (0) 2.4 (1.4) <.001Reduction at long-term follow-up (SD) 2.3 (0.8) 1.7 (1.5) .097

*Data available for 31 patients.Bolded numbers represent statistically significant P values.BMI, Body mass index; IQR, interquartile range; SBP, systolic blood pressure.

SurgeryVolume 156, Number 6

Wachtel et al 1397

Logistic regression analysis was performed onfactors identified on univariate analysis; contin-uous variables were dichotomized for ease of use.Age <50 years was not clinically relevant (oddsratio [OR] = 2.1, 95% CI = 0.63–6.8), but female

sex (OR = 14.3, 95%CI = 2.9–70.3), BMI#25 kg/m2

(OR = 6.0, 95% CI = 1.3–26.4), hypertension<5 years (OR = 5.3, 95% CI = 1.5–18.4), creati-nine #0.8 mg/dL (OR = 4.4, 95% CI = 1.2–16.3), and <2 preoperative antihypertensive

Table III. Factors associated with improved bloodpressure outcomes on long-term follow-up

Odds ratio

95%Confidenceinterval P value

Age <50 y* 2.1 0.63–6.8 .233Female sex 14.3 2.9–70.3 .001BMI #25 kg/m2 6.0 1.3–26.4 .019Duration of

hypertension <5 y5.3 1.5–18.4 .009

Serum creatinine<0.8 mg/dL

4.4 1.2–16.3 .028

<2 preoperativeantihypertensivemedications

9.3 2.5–34.7 .001

*Age <50 y was not statistically significant.Bolded numbers represent statistically significant P values.BMI, Body mass index.

SurgeryDecember 2014

1398 Wachtel et al

medications (OR = 9.3, 95% CI = 2.5–34.7) re-mained significantly associated with cure onlong-term follow-up (Table III).

The 5 statistically significant protective factors(female sex, BMI #25 kg/m2, hypertension<5 years, creatinine #0.8 mg/dL, and <2 preoper-ative antihypertensive medications) were incorpo-rated into a scoring system. A score of 0–1(n = 59) was associated with a 3.4% cure rate; ascore of 2–3 (n = 23) had a cure rate of 34.8%,and the few patients with a score of 4–5 (n = 3)had a 100% cure rate (Fig 1, A). When the ROCcurve was generated, the area under the curvewas 0.840, consistent with good fit to the data(Fig 1, B). A score of 4–5 was 23.1% sensitivityand 100% specific for cure at long-term follow-up; positive predictive value was 100%, and nega-tive predictive value was 87.8%.

Validation of ARS. The ARS was developed byZarnegar et al13 to evaluate outcomes 6months afteradrenalectomy for primary hyperaldosteronism. Itincorporates 4 protective factors (female sex, BMI#25 kg/m2, hypertension #6 years, and #2 preop-erative antihypertensivemedications) into a scoringsystem. Factors associated with cure in our patientpopulation overlapped with those used in the ARS,so we sought to compare the 2 scoring systems.

When theARSwas applied toourdataset, patientswith a score of 0–1 (n = 56) had a 1.8% cure rate;those with a score of 2–3 (n = 21) had a 28.6% curerate, and a score of 4–5 (n = 8) was associated witha 75.0% cure rate (Fig 2, A). For a score of 0–1, thenegative predictive value was 98.2%; for a score of4–5, the positive predictive value was 75.0%. Thearea under the ROC curve was 0.836 (Fig 2, B).

DISCUSSION

Hypertension persists in approximately 40–70%of patients at short-term follow-up after surgery forprimary hyperaldosteronism.10,12,14,15 Previousstudies have variously identified older age, longerduration of hypertension, increased serum creati-nine, use of >2 antihypertensive medicationspreoperatively, and positive family history of hyper-tension as factors predictive of persistent hyperten-sion at short-term follow-up.9,15 Correspondingly,older age and longer duration of hypertensionhave been associated with poor cardiovascular out-comes even after treatment of patients with primaryhyperaldosteronism due to either aldosteronomaor hyperplasia.16 Our results were consistent withthese findings. Factors protective of persistent hy-pertension in our cohort included female sex,BMI #25 kg/m2, hypertension <5 years’ duration,serum creatinine #0.8 mg/dL, and <2 preopera-tive antihypertensive medications. It is meaningfulthat factors predictive of short-term outcomes alsopredict long-term blood pressure control. Thesefindings suggest that patients who have resolutionof hypertension 6–12 months after surgery willlikely have durable blood pressure control andmay have less frequent follow-up. Conversely, pa-tients who have persistent hypertension 6–12 months postoperatively will likely benefit fromclose follow-up and careful medical management.

Persistent hypertension is likely secondary to acombination of concomitant essential hyperten-sion and permanent target organ damage. Severalstudies have demonstrated that despite completenormalization of plasma renin and aldosteronelevels in all patients after surgery, hypertensionpersists in a subset.14,17 The proportion of patientsaffected has a similar prevalence as essential hyper-tension in the general population, and patientshave similar risk profiles to those with essential hy-pertension. Although cardiovascular and renalcompromise have been shown to be partiallyreversible after adrenalectomy, longer duration ofmineralocorticoid excess appears to correlatewith more extensive impairment.18,19 Multiplemechanisms for irreversible target organ damageinclude hypertrophy of arterial intima and media,leading to increased vascular rigidity, myocardialfibrosis, and subsequent left ventricular hypertro-phy, and glomerular and tubuloinsterstitial injuryin the kidney.20-22

In our cohort, at short-term follow-up weobserved a cure rate of 6%, which increased to15% on long-term follow-up. Although short-termfollow-up data were limited, cure 6–12 months

Fig 1. Cure rates and performance of scoring system incorporating factors predictive of cure. (A) Cure rate by numberof protective factors, as identified in Table III. (B) Receiver operator characteristics (ROC) curve of scoring systemincorporating 5 protective factors. The value of the area under the curve was 0.840 (95% confidence interval 0.78–0.93). The 45-degree line represents the ROC curve for a model with a likelihood, which is the same as chance.

SurgeryVolume 156, Number 6

Wachtel et al 1399

postoperatively was 100% specific but only 50%sensitive for cure at long-term follow-up in thegroup with both short- and long-term follow-updata available. This may represent a slower returnto a physiologically normal state. This is supportedby the early decrease in antihypertensive agents weobserved, demonstrating that patients had sub-stantial improvement by 6–12 months postopera-tively but continued to experience gradualimprovement in blood pressure control over time.

Our cure rates were less than published curerates at short-term follow-up, which ranged from

31 to 87%.11,17 This large range reflects the varyingdefinitions of ‘‘cure’’ in the literature, with somestudies defining cure as normotension off all med-ications (the definition used in our study), whereasothers define cure as normotension regardless ofmedication status, or as decreased number of anti-hypertensive medications regardless of blood pres-sure control. Additionally, some older studies use avalue of #160/95 mm Hg as the definition ofnormotensive. However, compared with studies us-ing the same definition of cure, we still observed alower cure rate, which may reflect a greater

Fig 2. Cure rates and performance of the Aldosterone Resolution Score (ARS). (A) Cure rate by ARS. The ARS wasoriginally developed and validated on patients 6 months after adrenalectomy for primary hyperaldosteronism. Hereit is applied to a cohort of 85 patients at long-term follow-up. (B) Receiver operator characteristics (ROC) curve ofthe ARS. The value of the area under the curve was 0.836 (95% confidence interval 0.77–0.89). The 45-degree line rep-resents the ROC curve for a model with a likelihood which is the same as chance.

SurgeryDecember 2014

1400 Wachtel et al

prevalence of concomitant essential hypertensionand pre-existing target organ damage in our studypopulation. This theory is supported by the mark-edly greater proportion of high-risk patients (ARS0–1) in our cohort (68.2%, or 58 of 85 patients)compared with the combined cohort publishedby Zarnegar et al13 (43.7%, or 73 of 167 patients),or a separate validation cohort published by Ut-sumi et al23 (18.7%, or 17 of 91 patients).

The ARS is a 4-item scoring system that wasdeveloped and validated to evaluate outcomes6 months after adrenalectomy for primary

hyperaldosteronism. In our cohort it performedcomparably to our 5-item scoring system in termsof fit of the model. Although both models demon-strated excellent accuracy in our cohort, theincorporation of serum creatinine in our scoringsystem improves both the sensitivity and the pos-itive predictive value of a high score and mayprovide a valuable addition to identify patientslikely to be cured post-operatively. This remains tobe prospectively validated in a separate cohort.

In our study population, both cohortsbenefitted from adrenalectomy; there was no

SurgeryVolume 156, Number 6

Wachtel et al 1401

statistically significant difference in blood pressurereduction or in the decrease in number of antihy-pertensive medications between patients who werecured and those who were persistently hyperten-sive. This emphasizes the fact that virtually allpatients with primary hyperaldosteronism benefitfrom surgical intervention, which is consistent withfindings from other studies.10,12,17 These scoringsystems therefore should not prevent patientsfrom undergoing surgery but rather be used as apreoperative counseling tool to guide patient ex-pectations for long-term outcomes.

This study is limited by its retrospective natureand by reporting bias. Blood pressure measure-ments were obtained in 2 ways: measurement byclinicians recorded in the medical record andmeasurements by patients taken at home andreported to our research staff. A meta-analysis hasshown that self-blood pressure measurements arelower than those taken in clinic; some experts feelthat this reflects a lower incidence of ‘‘white coat’’hypertension.24,25 Additionally, self-reported mea-surements are subject to reporting bias. Althoughattempts were made to limit the inaccuracy of thedata by excluding patients with inexact or olderblood pressure measurements, this bias may affectoutcomes. Another limitation is the exclusion ofalmost half (48%) of patients who underwent adre-nalectomy, because of incomplete long-termfollow-up data. It is difficult to determine howthis may bias the cohort, because patients withpoor follow-up may have poorer blood pressurecontrol, or alternatively may have been cured andno longer require frequent follow-up. Within thecohort with long-term follow-up data, only 36%(n = 31) had short-term (6-month) follow-up dataavailable.

Previous studies have focused on short-termoutcomes after surgery for primary hyperaldoster-onism. In this study, we found that virtually allpatients benefited from surgery, and this benefitpersisted long-term. Factors associated with dura-ble blood pressure control after adrenalectomywere incorporated into a scoring system to predictlong-term outcomes after surgery. Compared withthe ARS, our scoring system performed as well ifnot better; the incorporation of serum creatininemay improve the ability to identify patients whowill be completely cured, although this remains tobe validated in a separate cohort. These data mayhelp guide preoperative patient counseling andexpectations for patients undergoing adrenalec-tomy for primary hyperaldosteronism.

REFERENCES

1. Douma S, Petidis K, Doumas M, Papaefthimiou P, Trianta-fyllou A, Kartali N, et al. Prevalence of primary hyperaldos-teronism in resistant hypertension: a retrospectiveobservational study. Lancet 2008;371:1921-6.

2. Eide IK, Torjesen PA, Drolsum A, Babovic A, Lilledahl NP.Low-renin status in therapy-resistant hypertension: a clueto efficient treatment. J Hypertens 2004;22:2217-26.

3. Tanabe A, Naruse M, Naruse K, Hase M, Yoshimoto T, Ta-naka M, et al. Left ventricular hypertrophy is more promi-nent in patients with primary aldosteronism than inpatients with other types of secondary hypertension. Hyper-tens Res 1997;20:85-90.

4. Strauch B, Petrak O, Wichterle D, Zelinka T, Holaj R, Wi-dimsky J Jr. Increased arterial wall stiffness in primary aldo-steronism in comparison with essential hypertension. Am JHypertens 2006;19:909-14.

5. Rossi GP, Bernini G, Desideri G, Fabris B, Ferri C, GiacchettiG, et al. Renal damage in primary aldosteronism: results ofthe PAPY Study. Hypertension 2006;48:232-8.

6. Milliez P, Girerd X, Plouin PF, Blacher J, Safar ME, MouradJJ. Evidence for an increased rate of cardiovascular events inpatients with primary aldosteronism. J Am Coll Cardiol2005;45:1243-8.

7. Savard S, Amar L, Plouin PF, Steichen O. Cardiovascularcomplications associated with primary aldosteronism: acontrolled cross-sectional study. Hypertension 2013;62:331-6.

8. Stowasser M, Sharman J, Leano R, Gordon RD, Ward G,Cowley D, et al. Evidence for abnormal left ventricular struc-ture and function in normotensive individuals with familialhyperaldosteronism type I. J Clin Endocrinol Metab 2005;90:5070-6.

9. Young WF. Primary aldosteronism: renaissance of a syn-drome. Clin Endocrinol 2007;66:607-18.

10. Sawka AM, Young WF, Thompson GB, Grant CS, Farley DR,Leibson C, et al. Primary aldosteronism: factors associatedwith normalization of blood pressure after surgery. AnnIntern Med 2001;135:258-61.

11. Carter Y, Roy M, Sippel RS, Chen H. Persistent hypertensionafter adrenalectomy for an aldosterone-producing ade-noma: weight as a critical prognostic factor for aldoste-rone’s lasting effect on the cardiac and vascular systems.J Surg Res 2012;177:241-7.

12. Celen O, O’Brien MJ, Melby JC, Beazley RM. Factors influ-encing outcome of surgery for primary aldosteronism. ArchSurg 1996;131:646-50.

13. Zarnegar R, YoungWF Jr, Lee J, Sweet MP, Kebebew E, FarleyDR, et al. The aldosteronoma resolution score: predictingcomplete resolution of hypertension after adrenalectomyfor aldosteronoma. Ann Surg 2008;247:511-8.

14. Proye CA, Mulliez EA, Carnaille BM, Lecomte-Houcke M,Decoulx M, Wemeau JL, et al. Essential hypertension: firstreason for persistent hypertension after unilateral adre-nalectomy for primary aldosteronism? Surgery 1998;124:1128-33.

15. Meyer A, Brabant G, Behrend M. Long-term follow-up afteradrenalectomy for primary aldosteronism. World J Surg2005;29:155-9.

16. Catena C, Colussi G, Nadalini E, Chiuch A, Baroselli S, La-penna R, et al. Cardiovascular outcomes in patients with pri-mary aldosteronism after treatment. Arch Intern Med 2008;168:80-5.

SurgeryDecember 2014

1402 Wachtel et al

17. Lumachi F, Ermani M, Basso SM, Armanini D, Iacobone M,Favia G. Long-term results of adrenalectomy in patientswith aldosterone-producing adenomas: multivariate analysisof factors affecting unresolved hypertension and review ofthe literature. Am Surg 2005;71:864-9.

18. Strauch B, Petrak O, Zelinka T, Wichterle D, Holaj R, Kasa-licky M, et al. Adrenalectomy improves arterial stiffness inprimary aldosteronism. Am J Hypertens 2008;21:1086-92.

19. Sechi LA, Novello M, Lapenna R, Baroselli S, Nadalini E,Colussi GL, et al. Long-term renal outcomes in patientswith primary aldosteronism. JAMA 2006;295:2638-45.

20. Widimsky J Jr, Strauch B, Petrak O, Rosa J, Somloova Z, Ze-linka T, et al. Vascular disturbances in primary aldoste-ronism: clinical evidence. Kidney Blood Press Res 2012;35:529-33.

21. Sechi LA, Colussi G, Di Fabio A, Catena C. Cardiovascularand renal damage in primary aldosteronism: outcomes aftertreatment. Am J Hypertens 2010;23:1253-60.

22. Marney AM, Brown NJ. Aldosterone and end-organ dam-age. Clin Sci (Lond) 2007;113:267-78.

23. Utsumi T, Kawamura K, Imamoto T, Kamiya N, Komiya A,Suzuki S, et al. High predictive accuracy of AldosteronomaResolution Score in Japanese patients with aldosterone-producing adenoma. Surgery 2012;151:437-43.

24. Appel LJ, Robinson KA, Guallar E, Erlinger T, Masood SO,Jehn M, et al. Utility of blood pressure monitoring outsideof the clinic setting. Evid Rep Technol Assess (Summ)2002:1-5.

25. Uhlig K, Balk EM, Patel K, Ip S, Kitsios GD, Obadan NO,et al. Self-measured blood pressure monitoring: compara-tive effectiveness. Rockville, MD: AHRQ; 2012.

DISCUSSIONDr Rebecca Sippel (Madison, WI): I don’t have

any disclosures. I think it’s nice to see these 2 pa-pers back to back, because it’s obviously showingthat there’s similar factors that predict resolution.I am a little concerned about the methodology inassessing the ARS. You developed this risk scorein this patient population, and, obviously, it’sgoing to perform better. It’s not really a fair com-parison. So I would just caution you on that. Itmight be nice since both of you have a cohort,that you could cross-validate each other’s models,and then we could really know which one is better.

Dr Heather Wachtel: The point is very welltaken. I think it speaks to the strength of these as-sociations that the factors that we independentlyidentified were virtually identical, and also isstrengthening to me that Dr Mitmaker’s groupfound kidney function to be related as well.

To address your second point, we actually havebeen in touch with Dr Zarnegar’s group and DrAronova, who just presented, and, hopefully, wewill have, in the future, a collaborative work topresent to you today.

Dr Tobias Carling (New Haven, CT): No disclo-sures. It’s unfortunate that we use the terms‘‘resolved’’ and ‘‘unresolved’’ hypertension, becausewhen we evaluate studies of outcome in patients

being operated for primary hyperaldosteronism,the fact that you don’t have a normalization of theblood pressure is not a sign of the operation not be-ing beneficial. It’s almost similar as saying that youdon’t have a normalized bone mineral density aftera parathyroidectomy that that would not be anappropriate operation. We know that the morbidityin primary aldosteronism is a lot more directly asso-ciated with aldosterone hypersecretion, as opposedto the elevated blood pressure by itself. One factorthat you obviously can’t measure preoperatively isthe genotype status of the underlying aldosterone-producing tumors. I wonder if you had the abilityto go back and look at the pathology specimenand see what proportion of patients had geneticmutations.

Dr Heather Wachtel: To address your first point,we absolutely agree with that. One of our concernswas that this type of scoring system might be usedto discourage patients from surgery. We do wish toemphasize that 98% of our patients had benefit,whether it was improvement in the number ofmedications or improvement in their BP control.Patients should absolutely not be deterred fromoperative intervention.

To address your second question, we have nothad the opportunity to look at this genotyping. Weare actually fortunate enough to have in ouraudience at this moment one of our extremelyillustrious colleagues from the pathology depart-ment. I hope that she’s listening to this and willhelp us with such a study in the future, Dr Livolsi.So that’s an important direction.

Dr Scott Wilhelm (Cleveland, OH): No disclo-sures. I found it very interesting that when youtalked about your patients who did not respond,their mean blood pressure decrease, I think, was24 millimeters of mercury and it was 23 in yourcure group. So your amount of change was thesame. Do you think that your starting blood pres-sure in the group that you considered failure wouldbe a reflection of potentially other causes of under-lying essential hypertension in addition to imposedprimary hyperaldosteronism on top of that?

Dr Heather Wachtel: Absolutely. We did look atthe levels of blood pressure preoperatively. Therewas no significant differences between the groupthat was cured and the group that was not. Howev-er, I think Dr Proye had published on your point inthe early 1990s. He mentioned that he believedthat underlying hypertension was the first poten-tial cause of persistent hypertension after surgeryfor primary hyperaldosteronism.

Dr Rasa Zarnegar (New York, NY): It’s an inter-esting topic. I have dealt with this for quite a long

SurgeryVolume 156, Number 6

Wachtel et al 1403

time, and I think we started off at the University ofCalifornia, San Francisco looking at this. I guessmy biggest issue with this study would be how it’sperformed, because you are really validating yourmodel. So basically your cure group only had 3 pa-tients with a 100% cure rate. Then when youcompare it with the University of California, SanFrancisco aldosterone resolution score, it pre-dicted 2 of those 3.

Therefore, I think you ought to be very cautiouswhen you say it’s a greater success rate in predict-ing cure when you are only looking at one extrapatient in your cure group. Because your curegroup is so small, in the 4–5 range, and you had 85patients, you are really limiting your scoring systemto a very small subpopulation of patients. Now, ifyou look at that, then that makes it very difficult touse it in the global system. What do you thinkabout that, and how would you address thatquestion as far as going forward with the study?

DrHeatherWachtel:We agree. Unfortunately, wewere limited a little bit in what we had to work with.We are looking, at this point, to obtain follow-updata from our entire cohort of 164 patients, whichshould be able to address this question.

The other alternative way to deal with this is tolook for validation in an external cohort. Hope-fully, we will be able to collaborate with some ofyou in the audience in the future.

Dr Quan-Yang Duh (San Francisco, CA): Veryquick question about the addition of creatininein your model. It actually worked both ways. Themore severe the disease of hyperaldosteronism,the greater your glomerular filtration rate (GFR).However, the more disease that you have of hyper-aldosteronism and hypertension can then give yourenal failure. So it can actually work in both direc-tions. So I want to hear your thought about yourexplanation of why the creatinine actually makesa difference in your model.

Dr Heather Wachtel: Our original hypothesiswas that a greater serum creatinine was a reflectionof a decreased kidney function, which was as aresult of longstanding hypertension, which hasbeen demonstrated to show effects on the glomer-ular apparatus as well as renal tubule or interstitialdamage. One of our transplant surgeons actuallyhad recommended that in a future prospectivestudy, that we attempt to biopsy every kidney, aswe do the adrenalectomy. I doubt that we aregoing to receive institutional review boardapproval to perform that study, but it would bean interesting way to get at the question.

Dr Quan-Yang Duh: In a cured patient, you actu-ally see a drop in GFR because the artificiallyincreased GFR in the diseased patient.

Dr Heather Wachtel: Thank you. I would love tohear your thoughts on that further.