Key words
primary aldosteronism - differential diagnosis - randomized controlled trial
Abbreviations
PA: Primary aldosteronism
CT: Computed tomography
AVS: Adrenal vein sampling
DDD: Defined daily dose
APA: Aldosterone-producing adenoma
MRA: Mineralocorticoid receptor antagonist
ADX: Adrenalectomy
Introduction
Primary aldosteronism (PA) represents the most frequent curable form of arterial hypertension
[1]
[2]
[3]. Its relevant cardiovascular comorbidities, with potential reversal or improvement
after early diagnosis and therapy, justify a screening in high-risk populations [4]
[5]
[6]
[7]
[8]
[9]
[10]. These include patients with resistant hypertension, spontaneous or diuretic-induced
hypokalemia, adrenal incidentaloma, or sleep apnea [11]. The two major forms of PA are unilateral aldosterone-producing adenoma (APA), which
is treated by surgery, and the more common bilateral adrenal hyperplasia, which is
addressed by mineralocorticoid receptor antagonist (MRA) treatment [12]
[13].
As therapeutic approaches diverge substantially between the two subtypes, differential
diagnosis is essential for patient care [14]
[15]. Following the establishment of the diagnosis of PA, current guidelines suggest
the performance of adrenal computed tomography (CT), which allows to exclude adrenocortical
carcinoma as a rare cause of aldosterone excess [11]
[16]. While CT scans have a high sensitivity in detecting larger adrenal tumors, they
only display a limited specificity for endocrine active lesions due to the high prevalence
of adrenal incidentalomas with increasing age [17]. Accordingly, bilateral adrenal vein sampling (AVS) is currently recommended as
the gold standard in differential diagnosis in PA patients willing to undergo adrenalectomy
(ADX) in case of unilateral disease. The rationale of this procedure is to measure
aldosterone directly at its offspring in the effluents of the adrenal veins. A systematic
review had pointed out a discordance between CT- and AVS-based diagnosis in approximately
40% of the cases [17]. While being safe in experienced hands, AVS has been judged as invasive, technically
demanding, and relatively expensive [18]
[19]. In many centers, success rates are low [20]. A relevant proportion of patients not treated in specialized centers is deprived
of this diagnostic tool due to lacking resources. Additionally, AVS is not well standardized
between centers regarding selectivity and lateralization indices, sequential or simultaneous
catheterization and the use of cosyntropin, thereby complicating comparability of
diagnosis findings. Finally, it has been criticized, that prospective studies for
the reliability of AVS results are lacking. As a consequence, efforts have been made
to avoid AVS at least in some patients based on different algorithms and scores that
had been demonstrated to provide some level of subtype prediction [21]
[22]
[23]. Alternative functional imaging techniques including metomidate PET-CT or the use
of specific aldosterone synthase tracer are currently under investigation [24]
[25].
In this clinical scenario, the SPARCTACUS trial (Subtyping Primary Aldosteronism:
A Randomized Trial Comparing Adrenal Vein Sampling and Computed Tomography Scan) recently
reported by Dekkers and colleagues in the Lancet Diabetes & Endocrinology, set out
to compare AVS-based and CT-based treatment outcome in PA patients [26]. The study reported that treatment decisions based on both diagnostic tools led
to similar blood pressure improvement and health-related quality of life in PA patients
at one year of follow-up. Therefore, the authors postulated that the extra-costs of
AVS were dispensable and that neither AVS nor CT would correctly predict PA subtypes
in all cases. Due to its unexpected findings with potentially relevant and practical
consequences for patient care, publication of the study triggered extensive discussions.
The current review aims to outline the different points of view regarding the protocol
and the results of the study.
Protocol PRO
The primary endpoint of the SPARCTACUS study was defined as blood pressure control
at one year of follow-up [26]. Indeed, the aim to improve clinical outcome represents the real gold standard of
diagnostic or therapeutic procedures. As AVS is being criticized to be unavailable
for the majority of PA patients due to its complexity and restriction to specialized
centers, simplified algorithms with improved accessibility are required. The implementation
of CT scanning instead of AVS, at least in a certain proportion of PA patients, would
ease the diagnostic work-up in this patient population.
For the completion of this goal, it is important to provide clinical paths that are
based on the best scientific evidence, that is, randomized controlled trials. In fact,
the SPARTACUS study followed a diagnostic, randomized, controlled, multicenter design.
PA was confirmed by accepted clinical standards including a salt-loading test. The
sample size was determined prior to the start of the study by power calculations:
200 patients with PA were randomly assigned using a web-based algorithm to receive
AVS (preceded by CT) or only CT to subtype PA. Randomization resulted in a well-balanced
distribution of patients between the two groups. Patients randomized to CT were adrenalectomized
in case of a unilaterally enlarged adrenal gland and a normal appearing contralateral
gland. In contrast, patients with normal or bilaterally enlarged adrenal glands were
treated with mineralocorticoid antagonists. For better comparison, adrenal CT was
assessed by a local radiologist and reviewed by a central facility. In cases of discrepancy
between these CT readings, the final decision was taken by the local center.
AVS performance is variable across centers in term of procedure and interpretation
[27]
[28]. In the SPARTACUS study, AVS was performed according to accepted protocols under
continuous cosyntropin stimulation with sequential catheterization of the adrenal
veins. Based on a selectivity index of 3.0 or higher, the success rate for bilateral
cannulation of the adrenal veins was reported as high as 96%. Patients with unilateral
disease (based on a lateralization index of 4.0 or higher and a suppression index
of 1.0 or lower) were adrenalectomized. In cases of bilateral disease, mineralocorticoid
antagonist treatment was initiated. In those instances when AVS failed, patients were
treated according to CT findings. Thereby, the applied criteria for the interpretation
can be regarded as strict enough to ensure rigorous diagnosis. In two large studies,
ACTH stimulation resulted in 1–4/46 difference in diagnosis compared with unstimulated
procedures [29]
[30]. Therefore, it is highly improbable that this could have an impact on the final
outcome of the study.
From initially 200 enrolled patients, 184 completed follow-up with even distribution
in the AVS and CT group. Patients were investigated according to an intention-to-diagnose
analysis for the primary endpoint at 12 months following therapy. In fact, this interval
has recently been endorsed by an international expert panel as a relevant time point
for re-assessment of clinical outcome in PA patients [31]. As a measure of blood pressure control, the intensity of antihypertensive treatment
for obtaining target blood pressure (<135/85 mmHg using a semiautomatic device or
<140/90 mmHg using office measurement) was quantified as defined daily dose (DDD).
This approach represents a practical endpoint, as blood pressure is the parameter
relevant for the patients [32]. Ambulatory blood pressure was used, which is a very objective form of monitoring.
Key secondary endpoints were biochemical outcome in patients who had undergone ADX,
which was analyzed by salt-loading test. Further endpoints included physical and mental
scores, the proportion of patients reaching target blood pressure, adverse events,
and cost-effectiveness. The latter point is of great importance, as AVS is relatively
expensive, which might not be justified if it would lack diagnostic superiority.
In summary, patients were scrutinized by a well-defined clinical protocol, allocated
into diagnostic procedure groups in a randomized fashion and prospectively followed
up. In this regard, the SPARTACUS trial has implemented the highest standards of a
clinical study design. To this point, it is the first and only in the field of clinical
PA research that aims at this evidence level.
Protocol CONTRA
The SPARTACUS trial randomized 184 subjects with PA to adrenal CT scanning or adrenal
CT scanning plus AVS to establish the final subtype diagnosis and address the patients
with a unilateral disease to surgery [26]. The authors should be commended for the performance of the first randomized trial
on this disease, however the design of the SPARTACUS study has some weak points that
could reduce the relevance of the results and their application in clinical practice.
The most important points are: 1) DDD is not the appropriate primary end-point for
this type of study; 2) the SPARTACUS cohort is not representative of the general PA
population and therefore the results cannot be generalized; 3) the comparison between
AVS versus CT-based MRA therapy was not necessary and reduced the power of the study;
4) the sample size of the investigated population was not adequate to prove the non-inferiority
of CT scanning in comparison with AVS to determine indication for ADX.
The first consideration is that DDD is not the appropriate primary end-point. First,
it does not evaluate the biochemical cure of PA, the most reliable measure of the
success of ADX [31]; second, it does not take into account the concomitant presence of essential hypertension
that can confound the final outcome judgment; further, the intensity of the treatment
depends largely on the type of drug that is considered in the calculation. As an example,
a patient with blood pressure levels of 130/80 mmHg before ADX under spironolactone
75 mg, amlodipine 5 mg, lisinopril 10 mg, and hydrochlorothiazide 12.5 mg has a DDD=3.5;
the same patient, with the same blood pressure after ADX, taking ramipril 10 mg as
monotherapy, has a DDD=4. Clinicians would unanimously consider this patient largely
improved and not worsened as it appears if the DDD is used as an indicator of clinical
success after unilateral ADX.
The second consideration is that the SPARTACUS cohort is not representative of the
general PA population and, therefore, the results cannot be generalized. Criteria
for inclusion were: hypertension requiring 3 or more antihypertensive drugs in adequate
doses and/or hypertension accompanied by spontaneous or diuretic-induced hypokalemia
(serum potassium <3.5 mmol/l), which means that only patients with a severe phenotype
of PA were selected. In the recent PATO (primary aldosteronism in Torino) study, performed
on the general hypertensive population seen in primary care practice, the majority
of PA patients would not have been included in the SPARTACUS trial because of a milder
phenotype [33]. For example, in the PATO study, APA accounted for 25% of all PA cases, whereas
in the SPARTACUS their prevalence was 50%; furthermore, hypokalemia was observed in
29% of the patients with PA in the PATO study versus 68% in the SPARTACUS cohort.
The data of the PATO study are also coherent with a retrospective evaluation of the
clinical and biochemical features of patients with PA in referral centers from five
continents [2]. It is conceivable that patients with a severe PA phenotype and high prevalence
of APA will respond to ADX (even if AVS is not performed), but this would not be the
case for patients with a mild form of the disease.
Another major flaw of the design of the study is the choice of randomizing patients
to CT-based MRA therapy and AVS-based MRA therapy [26]. It is in fact well known, since the publication of the studies performed by the
Cleveland Clinic group, that patients with APA respond well to MRA [34]. Similarly, a more recent study demonstrated that adequate doses of spironolactone
determine a blood pressure reduction similar to that obtained with ADX [35]. Therefore, this arm of the study reduces the power of the other comparison, without
providing any useful novel information.
In a systematic review/meta-analysis, Kempers et al. showed that AVS and CT scanning
result in a different diagnosis in around 38% of cases [17]. In 19% of cases, patients would have been inappropriately excluded from ADX following
CT scanning, where AVS showed unilateral secretion; however, this discrepancy would
not affect the results of the SPARTACUS study. Therefore, only 18.5% of patients (14.6%
having an inappropriate ADX when AVS showed a bilateral disease plus 3.9% having ADX
on the wrong side when AVS showed aldosterone secretion on the opposite side) would
be inappropriately adrenalectomized following CT scanning instead of AVS. It should
be emphasized that ADX is also effective in some selected patients with bilateral
PA: in fact, it resulted in hypertension cure in 15% of cases and improvement in another
20% [36]. Based on this assumption, no more than 6 to 9 patients are expected to have persistence
of PA if a cohort of 46 patients is adrenalectomized following CT scanning alone;
interestingly, the persistence of PA in the SPARTACUS study is 9/46.
In the primary aldosteronism surgery outcome (PASO) study, AVS resulted in a complete
biochemical cure of PA in 94% of the patients [31]. Using the expected rate of cure of ADX following the indication of CT scanning
versus AVS, a number of 258 patients would have been necessary instead of 46, to prove
the non-inferiority of CT scanning with respect to AVS for the diagnosis of unilateral
PA.
In conclusion, the SPARTACUS trial conveys the strong message that in patients with
PA ADX based on CT diagnosis has a similar outcome compared with ADX based on AVS
findings, thereby challenging the current Endocrine Society Guideline [11]. However, the above discussed pitfalls significantly affected the results and limited
its generalizability to the whole PA population.
Results PRO
Although the clinical, cardiac, and renal outcomes of PA have been shown to be comparable
in patients treated with ADX or MRA [37], differentiation between unilateral or bilateral forms of this condition is still
widely considered to be essential for definition of the appropriate therapeutic choice
[11]. To this purpose, different approaches have been used in the past, including CT
or MRI-based imaging, adrenal scintigraphy, metomidate PET-CT, and AVS. Previous retrospective
investigations pointed out a substantial discordance (more than 40%) between AVS and
CT in differentiation of unilateral from bilateral adrenal disease in PA [17]. Because of the functional information provided by AVS, this was indicated as the
“gold standard” for differentiation, thereby generating the preconception that AVS
is almost always right, whereas CT is frequently wrong. As a consequence, AVS has
been asserted as the unavoidable cross road in the diagnostic workup recommended to
the majority of patients with PA [11]. However, no demonstration of this alleged superiority of AVS over other diagnostic
methods for characterization of unilateral or bilateral forms of PA could be found
in the medical literature and until the publication of the SPARTACUS study [26] no prospective assessment of this issue had been done. As already stated, randomized
controlled trials provide the best clinical evidence for clinical decisions and the
SPARTACUS trial was the first of this kind. The study compared the outcome of CT-based
management with AVS-based management in an appropriately sized sample of patients
with PA who were treated with either ADX or MRA and were followed for one year. The
outcome was assessed in a intention-to-diagnose analysis and both in primary (DDD
and number of antihypertensive drugs used at follow-up) and most of secondary endpoints
(proportion of patients reaching target blood pressure; serum potassium; plasma aldosterone
levels after salt-loading post-ADX; patients with biochemical evidence of resolved
PA; health-related quality of life, physical and mental; adverse events) no significant
differences were observed. The only difference was found in the mean total cost of
the procedure per patient that was 60% higher in those patients who underwent AVS.
Notably, primary and secondary endpoints did not differ between the CT and AVS group
even when patients who were treated with surgery or MRA were analyzed separately.
Although some reasonable and also some definitely questionable critiques have been
raised to the SPARTACUS study, it cannot be denied that this is the only study that
has approached the issue of the validity of AVS in a prospective randomized protocol,
providing a clear demonstration that if CT is not foolproof for differentiation of
unilateral from bilateral forms of PA, AVS is no better. These conclusions wipe out
the misconception that AVS could be considered a gold standard for definition of subtypes
of PA and undermine the Manichaean view that many have of it.
The results of the SPARTACUS study should not get to surprise because data on AVS
previously obtained in the top referral centers performing AVS worldwide had already
pointed out at its serious limitations. In the German’s Conn registry the results
of 200 AVS procedures were analyzed in two phases, retrospective and prospective,
after introduction of measures designed to improve the rates of successful cannulation.
The rate of success in correct collection of adrenal samples was less than one third
in the retrospective phase and less than two thirds in the prospective phase [20]. Also, the rate of success was extremely variable, from 80% to less than 30%, depending
upon the stringency of the selectivity index that had been used. Similar findings
were reported in Turin where the results of AVS in 64 patients with PA who had undergone
the procedure twice showed an impressive disparity in the definition of successful
cannulation of adrenal veins and lateralized aldosterone secretion depending upon
the stringency of the criteria that were used [38]. In this study, the rate of concordance among three different criteria used for
definition of lateralized secretion was 32% and the rate of concordance between the
two procedures performed in the same patient was 35%. To notice, lateralization as
detected by AVS changed from unilateral in one side to unilateral in the contralateral
side in 14% of patients. In Paris, more than 500 AVS were retrospectively reviewed
comparing the different diagnostic criteria used in 4 of the top referral labs for
AVS [39]. Comparison between the lab that used the most stringent criteria with the lab that
used the most lenient showed a five-fold difference in the proportion of unsuccessful
procedures (18% vs. 4%) and a two-fold difference in the proportion of lateralized
aldosterone secretion (26% vs. 60%). Because of the lack of standardization of AVS
among referral centers consensus documents have been published by expert committees
[27]
[28], but if one looks carefully at them many substantial differences still can be found
showing that there is very little consensus even among experts. Thus, while being
relatively safe in experienced hands, the procedure of AVS is invasive, technically
demanding, relatively expensive, and inadequately standardized and, in the light of
the findings of the SPARTACUS study, does not seem to offer any advantage over CT
in the outcome of patients treated for PA.
In summary, the findings of the SPARTACUS study strongly support the concept that
AVS is not a gold standard for differentiation of PA subtypes and keep wide open the
possibility to define the opportunity for unilateral ADX or, alternatively, MRA treatment
with diagnostic approaches, such as CT, that have a comparable level of reliability.
Needless to say, this should be always done under the guidance of a balanced clinical
judgment that takes into account all information on each single patients that is the
first ingredient, beyond and above guidelines, for taking appropriate clinical decisions.
Results CONTRA
The investigators for the SPARTACUS Trial are to be congratulated for completing the
first prospective study comparing CT- versus AVS-guided treatment of patients with
PA, an impressive achievement given the large amount of planning, workload and funding
support that would have been required. However, careful examination of the results
reveals a number of anomalous findings that raise serious concerns about the validity
and generalizability of the data and the conclusions that have been drawn. Furthermore,
there are clear trends towards superiority of AVS that add weight to the argument
provided above that the power of the study was insufficient to show significant differences
between the two study groups in terms of treatment (and in particular, ADX) outcomes.
There are a number “odd” findings in SPARTACUS:
-
The rate of lateralization in the CT and AVS groups was exactly the same at 50%, which
is in sharp contradiction to the reports of previous studies in which centers relying
on CT-based subtype differentiation found much lower rates of detection of APA than
those employing AVS [2]. One potential explanation for this is the very permissive criteria used for lateralization
on CT, requiring only an enlargement (not even a mass lesion) of an adrenal, defined
as a thickness of 7 mm or more in the body or limb. Surely the investigators are not
suggesting that, with all we have learned through countless previous studies about
the unreliability of even a mass lesion on CT, this is sufficient to warrant proceeding
to surgery without any other supporting evidence of lateralization whatsoever? Another
possible explanation for this anomalous result is selection bias towards subjects
with more florid (and hence more likely unilateral) varieties of PA, leading to an
over-representation of patients with larger APAs more easily detectable by CT, which
is supported by other lines of evidence outlined below.
-
A 50% rate of lateralization is high even for AVS-based subtype differentiation when
compared with most other recent reports [2]
[40] with the exception of some Asian cohorts [41]
[42], and centers which use very permissive lateralization criteria [38]
[43] (which the SPARTACUS investigators did not). This again suggests selection bias.
-
Moreover, the high proportion of hypokalemic patients (>60%, where most other studies
report hypokalemia in the minority) among the SPARTACUS cohort is further evidence
for selection of more severe, advanced PA, which may have been easier to localize
by CT.
-
Whereas most PA cohorts show a roughly equal gender distribution, in SPARTACUS, males
made up over three-quarters of PA patients. This inexplicable result again hinders
generalization of SPARTACUS findings to other centers.
-
There was a much higher localization of “APA” to the left (76% vs. right 24%) in the
CT group compared to the AVS group (54 vs. 46%) and to PA cohorts in other studies.
This may, at least in part, be due to nearby splenic vessels which can be mistaken
for APAs. Whatever the reason, it does not bode well for the validity of CT lateralization.
-
The overall very low rate of HT cure (14 of 92=15%) among the ADX patients is also
unexplained. This mirrors the relatively poor outcomes of the same Netherlands and
Polish centers that contributed to the recently reported PASO study when compared
with almost every other center that participated [31]. Not only does this raise further concerns about generalizability, but it also would
have seriously impacted on the power of the study to show differences in cure rates
between the CT- and AVS-based treatment groups (see below).
-
Given that the real “proof of the pudding” in terms of attempts at lateralization
is in the response to ADX, it is uncertain why the investigators even bothered to
look at responses to MRA. But even there, the surprising finding that non-lateralizing
patients in the AVS group need more DDDs than those in the CT group (median 5.7 vs.
4.0; p=0.05) defies logical explanation and casts doubt on the effectiveness of randomization.
Notwithstanding the many anomalous findings and concerns about selection bias and
generalizability raised above, there were still several important findings in SPARTACUS
that argued against a CT-based approach but which were largely left unmentioned by
the authors:
-
In the CT-based treatment group, consensus could not be reached regarding lateralization
in a sizable proportion of patients (11 of 98), almost all of whom were assigned to
MRA, whereas AVS was unsuccessful in permitting a diagnosis in only four of 96.
-
In keeping with an enormous body of existing data, a full 50% of the 90 patients from
the AVS group who had both conclusive CT and AVS demonstrated discordant results between
the two procedures.
-
Most importantly, despite the very low hypertension cure rate observed following ADX,
there was still a strong trend towards a superior cure rate among the AVS-based treatment
group (22%) compared with the CT group (9%) which almost reached statistical significance
(p=0.08). Had the study been powered to examine this (rather than DDDs) as the primary
endpoint, as has been recommended by the PASO investigators [31], it is highly likely that a significant difference would have been observed. As
it is, with such a low overall rate of hypertension cure, SPARTACUS was clearly seriously
underpowered.
-
Biochemical responses to ADX also tended towards superior outcomes for the AVS group,
with persistent PA being observed in almost double the operated patients in the CT
group compared with the AVS group (20 vs. 11%), but again with numbers too small to
reach statistical significance.
In short, SPARTACUS results are non-generalizable and the study was powered to the
wrong primary endpoint (rather than more meaningful ones such as cure of hypertension
and PA in response to ADX). Despite this, the trends for superiority of AVS were clearly
there but unfortunately ignored.
Even if we give SPARTACUS the benefit of the doubt and accept that removing the wrong
gland, or inappropriately removing a gland from a patient with non-lateralizing PA,
occurs in only a minority of patients in whom management is guided by CT (and was
therefore not detectable in this underpowered, cohort-based study), such undesirable
outcomes should be avoided wherever possible.
Conclusions
SPARTACUS has attempted to address an important clinical question in comparing AVS-
with CT-based decision making in terms of the outcome of surgical and specific medical
treatment in patients with PA. Its strengths include its robust protocol, the fact
that it is the first randomized, prospective study in its field, the relatively strict
criteria used to determine lateralization of aldosterone production on AVS, and the
use of 24-h ambulatory blood pressure monitoring to assess blood pressure outcomes.
However, there are also significant limitations, the most concerning of which are
evidence of selection bias, anomalous results, the unusual choice of primary endpoint
(DDDs), the decision to include responses to MRAs, and the low power of the study
to show difference in more traditional endpoints (particularly cure of hypertension
post ADX). Hence, while both CT and AVS are clearly imperfect in predicting responses
to PA treatment, it remains debatable as to whether SPARTACUS has managed to definitively
refute the long held view that AVS is still best.
