Somatic Mutations in APA
The most exciting finding over the past 5 years has been that of the characterisation
of somatic (i. e., confined to the adrenal) mutations in over 50% of aldosterone
producing adenomas (APA). The initial, prismatic report by Choi et al. [3] from the Lifton laboratory found somatic
mutations in KCNJ5, encoding the Kir3.4 potassium channel, in 8 of 22 APA from
Europe and the USA. In this series, the mutation-bearing APA were larger, more
common in females, found at a younger age than wild-type APA, and apparently more
florid, reflected in higher aldosterone/renin ratios (ARR).
Shortly thereafter, in a much larger collaborative study from the ENS@T Group in
Europe, these findings were confirmed in large part and extended [4] in over 300 patients. The percentage of
mutation bearing tumours was similar. The female predominance was similarly
pronounced, and they were similarly more florid on the basis of higher plasma
aldosterone concentrations (PAC). Points of difference were in the proportion of the
2 mutations found: in the Choi study of 8 cases, 2 were G151R and 6 L168R; in the
ENSAT study the numbers were reversed to 76 G151R vs. 53 L168R. Secondly, in the
larger study, no difference in size was found between mutation-bearing and wild-type
APA.
Subsequently, the area has been rapidly extended in several ways. First, and
successively, other genes have been found in APA – for sodium and calcium ATPases
(ATP1A1, ATP2B3) in 2 different series [5]
[6], and in the calcium channels
CACNA1D [7] and CTNNB1 [8]. Although the proportions of these mutations
vary between laboratories (e. g., ATPases: [5]), those in KCNJ5 remain the most common to date. In a very recent series
[8], the breakdown was KCNJ5 37, CACNA1D:
10, ATP1A1: 8, ATP2B3: 3, and CTNNB1: 2. Perhaps of interest in terms of the genesis
of APA, CACNA1D mutations were found in 2 hyperplastic adrenals without an adenoma.
Whereas the KCNJ5 mutation-bearing APA are more prevalent in females, the opposite
appears to be the case for the others as a whole. Although for the less common
variants numbers are smaller, they are also on average less florid in phenotype.
The second extension has been to document the variety of KCNJ5 mutations in addition
to the 2 initially described [3]
[4], the extent to which the mutation increases
intracellular calcium concentrations, and their phenotype: to date more than half
a
dozen different loci have been reported [9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]. Third, a series of
relatively small studies, in most cases, have extended the countries of origin of
the original cohorts (Europe, America) to Japan and China [17]
[18]
[19]
[20]
[21]
[22]. In both the latter the
percentage of mutation-bearing APA is commonly reported to be much higher than in
European, American, and Australian cohorts with APA. This has led to speculation
about ethnic differences: a more likely reason for the higher percentage is
selection bias, until proven otherwise.
Intercalated Cells
In addition to their ground-breaking discovery of KCNJ5 somatic mutations in APA,
the
Lifton laboratory subsequently published a breathtakingly elegant study documenting
a hitherto undescribed additional pathway of distal tubular sodium retention.
Current orthodoxy locates mineralocorticoid receptors (MR) in principal cells in the
distal tubule, where they are rendered aldosterone selective (‘protected’) via the
action of the enzyme 11βHSD2. The paper by Shibata et al. [32] demonstrates the presence of MR but not
11βHSD2 in intercalated cells. In the resting sodium replete state, these MR are
phosphorylated on serine 843, and thus rendered inactive. When, however, levels of
angiotensin rise in response to volume depletion or sodium deficiency, angiotensin
acts through a WNK4 signalling pathway to dephosphorylate MR, allowing them to be
activated, by either aldosterone or cortisol, consistent with the absence of
11βHSD2.
What the ramifications of this discovery are for the pathophysiology of primary
aldosteronism are not clear. They provide a satisfying explanation for the
difference between the MR–/– and the aldosterone synthase –/– mouse. The former
cannot survive a low sodium diet; the latter shows a fall in blood pressure,
unsurprisingly, but survives – presumably due to the high angiotensin levels acting
in intercalated cells to dephosphorylate MR S843, which is then activated
by corticosterone to retain sodium [33]. Given
that we are yet to understand the pathogenetic mechanisms of blood pressure
elevation and tissue damage in primary aldosteronism, studies on angiotensinogen –/–
mice might also be worth a moments’ reflection: in response to sodium deprivation
–/– and wild-type mice elevate their PAC to indistinguishable levels [34].
Net of these findings, there are no new answers over the past 5 years: other advances
have been incremental rather than paradigm shifts. Many of these in fact represent
new questions, in that they constitute a point of departure, preliminary studies
that need to be repeated, extended and able to be generalised before being admitted
to the canon of ‘new answers’, hopefully by 2020. If this is the case, then by that
time we may see real progress in the management of primary aldosteronism.
New Answers: Basic Studies
First, we may see major advances in the pathophysiology of both APA and BAH; there
is, for example, tantalizing preliminary evidence for the drivers of bilateral
adrenal hyperplasia. At present, over 50% of patients with an APA can be shown to
have a mutation in any one of 5 different genes: anything is possible, but it seems
very unlikely that all of the remainder are truly ‘wild-type’. We know how the
presently detected somatic mutations increase aldosterone secretion, but not the
steps in adenoma formation. Zona fasciculata cells appear to harbour the KCNJ5
mutation, but APA with the other somatic mutations are of mixed cell type: what does
this imply, for adenoma formation, for oversecretion of aldosterone?
With the advent of specific antisera able to discriminate human CYP11B1 from CYP11B2
the histology of the adrenal cortex has been established with a precision not
previously possible on the basis of cell morphology on haematoxylin and eosin
stained sections [35]. Counterintuitively,
CYP11B2 expression has been reported as essentially absent in some APA [36], but present in aldosterone producing cell
clusters in normal adrenals and those from patients with primary aldosteronism. Very
recently, in normal adrenals from renal transplantation donors, next generation
sequencing revealed that in 8 of 23 cases the APCC harboured aldosterone driver
mutations, absent from the surrounding zona glomerulosa and with a pattern of
mutation (6 CACNA1D, 2 ATP1A1) different from that found to date in mutation-bearing
APA [37]. This is a prismatic finding, and
bodes well for major advances over the next 5 years in terms of the preconditions
for pathophysiological aldosterone excess.
Of one thing we can be certain, that whether or not there are answers to these
questions in 5 years time it will not be for want of trying.
New Answers: Clinical Questions
1. Confirmatory testing
In the clinical area things may unfortunately move much more slowly: of all the
possible clinical advances this essay will consider 5 in detail. The first,
simplest but perhaps the least tractable is the existence of 5 or more
confirmatory/exclusion tests used to winnow ‘true’ PA from within the group of
hypertensives who screen positive for PA on the basis of an elevated aldosterone
to renin ratio (ARR). Direct comparisons between tests are few, often post-hoc,
on relatively small groups of subjects. It is, of course, possible that no one
test is superior to another – or any other – of the 5, in terms of specificity
and selectivity, possible but unlikely. Currently what maintains the diversity
may be lack of evidence that a particular test is inferior, but is more likely
to be habit-formed by local history and usage, familiarity and – often
underestimated – the desire not to compromise a series by changing horses in
mid-stream.
Among the confirmatory/exclusion tests the fludrocortisone saline suppression
test (FST) is sometimes (disputedly) referred to as ‘the gold standard’. In
terms of hospitalisation (not mandatory), time, compliance, adequacy of sodium
and potassium intake and cost it is clearly the most complex. To their credit
the Brisbane group recently and very successfully compared their customary FST
with a seated saline suppression test (SSST), distinct from the customary
recumbent oral or intravenous saline suppression tests [38]. In the preliminary study they reported
a 23/24 patient congruence between the FST and the SSST, compared with much
lower levels between the FST and recumbent saline suppression tests. There is in
progress a larger study: if this recapitulates the preliminary study, there
would appear to be a very strong case for its widespread adoption, and
discontinuation of the current range of testing, on the basis of accuracy,
convenience, compliance and cost. It would, however, be unwise to hold your
breath until this happens.
2. Assays, ‘normal’ levels, and cut-offs
A second, not dissimilar area of variation is in the measurement of renin and
aldosterone, the overly ample ‘normal ranges’ for plasma aldosterone, with the
result – at least in part – of variation in cut-offs for a positive ARR and
elevated plasma aldosterone levels. Immunoassays of aldosterone are notoriously
unreliable; renin activity or concentration are at least one step backwards from
angiotensin II. The goal therefore in terms of measurement is to measure
aldosterone by mass spectrometry, and similarly plasma angiotensin – or a
validated surrogate. This means that proposed aldosterone to angiotensin ratio
measurement needs to be centralised into tertiary medical institutions, rather
than the ARR being done at present in a wide-ranging variety of laboratories.
Importantly, the normal ranges of plasma aldosterone concentration need to be
revisited, with due regard to ethnic variation and differences in sodium and
potassium intake. The published normal PAC at the Mayo Clinic has been
4–21 ng/dl, and that in Ancona 3–31 ng/dl: both ranges are unlikely to reflect
ethnic differences, nor differences in dietary intake.
3. Avoiding AVS
A third area in which answers are needed is that of alternatives to adrenal
venous sampling (AVS) as a means to lateralise (or not) the source of the
hyperaldosteronism. In patients with hypokalaemia, immeasurably low renin and
high aldosterone, some centres omit confirmatory/exclusion testing, and move
directly to AVS. In others, similar young patients (< 35–40) with a
unilateral adenoma and a contralateral normal adrenal on imaging have proceeded
to surgery without AVS, on the basis of the unlikelihood of adrenal
incidentaloma at that age. There have been several post hoc studies on adrenal
venous and peripheral levels of 18-hydroxy- and 18-oxocortisol in patients
previously lateralised on AVS, as possible predictors of a distinction between
high levels (APA) and low levels (BAH). In 2 such studies [39]
[40] there were highly significant differences, with the latter
claiming ~ 60% ability to discriminate on the basis of peripheral levels, so
that only ~ 40% of patients would require AVS. Given the cut-offs widely but not
universally used in Japan, these values may or may not be generalisable
elsewhere. Finally, if APA cells shed into the bloodstream can be confirmed as
harbouring one of the somatic mutations described above, and imaging is
consistent with a clearly unilateral adenoma, it would appear appropriate to
omit AVS.
4. Low Renin Hypertension?
The fourth area is that of whether or not the current cut-offs, and subsequent
workup, miss a major group of patients with primary aldosteronism. This question
was first addressed over 35 years ago [41], and answered positively. Helber and colleagues took 3 groups – 35
healthy normotensive controls, 100 ‘essential hypertensives’ and 16 patients
with proven primary aldosteronism (12 APA, 4 BAH), maintained them on a fixed
high sodium intake, and measured urinary aldosterone levels, plasma electrolytes
and blood pressure. What they found was unequivocal, despite which it has been
largely ignored. All but 2 (of the 4 BAH) the proven primary aldosteronism
patients had urinary aldosterone levels above the normal range. Among the 100
‘essential hypertensives’, those with normal aldosterone levels (group A, n=64)
had average urinary aldosterone levels of 2.7±1.4 μg/day, compared with
10.0±3.0 μg/day in group B (n=36) who had levels above the normal range. In
group A the mean plasma [K+] was 4.26 meq/l; in group B 3.81 meq/l.
Tellingly, when all patients were put on a trial of spironolactone, in group A
mean blood pressure fell by 9 mm Hg, compared with 21 mm Hg in Group B. On these
findings it seems that at least 36% of hypertensives in this cohort had occult
primary aldosteronism, given that those with established PA had already been so
diagnosed. This, of course, was in stark opposition to received wisdom at that
time, that primary aldosteronism was a rare (< 1% of all hypertension) and
relatively benign form of elevated blood pressure.
Over 20 years later a study on the role of MR, rather than aldosterone per se,
was published [42]. In this study 397
essential hypertensives, carefully screened and normal in terms of everything
except their starting BP (average value: 154/100 mm Hg) were enrolled in a
titration-to-effect trial of eplerenone as monotherapy. For the first 4 weeks
all subjects received 50 mg/d eplerenone, at the end of which period 44% reached
goal BP (diastolic<90 mm Hg) with a marked fall in BP (16/14); they continued
on 50 mg and the non-responders (BP fall 6/3) went onto 100 mg/day for weeks
5–8. At 8 weeks 66/156 of the non-responders to 50 mg/day showed a mean fall in
BP of 20/15 mm Hg, compared with a fall of 5/3 mm Hg in those who did not reach
goal BP on this dose. The responders to 100 mg/day continued on that dose, and
the 156 non-responders were given 200 mg/day for weeks 8–12. On this dose 75
subjects reached goal DBP, with a mean fall of 17/15: the remaining 81 subjects
(~ 20% of the total group) were unresponsive (2/1) to mineralocorticoid receptor
antagonist administration.
The interpretation of these data was that in ~ 80% of essential hypertensives the
elevation in blood pressure is dependent at least in part on MR activation.
Given the screening process for entry into the study (mean plasma
[K+] was 4.27 meq/l, cf. 4.26 meq/l in the Helber et al. study cited
above) it is difficult to ascribe MR activation to aldosterone. On the other
hand, normal levels of cortisol occupy the majority of MR in 11βHSD2 protected
tissues as an aldosterone antagonist under normal circumstances; in the context
of tissue damage, however, as occurs in hypertensive vasculature, cortisol
becomes an MR agonist [43]. That the
effect of eplerenone is not due to a diuretic action is shown by the finding
that at no dose did the obligate changes in plasma [K+] differ
between responders and non-responders, underlined by the title of the paper –
“Distinguishing the Blood Pressure and Electrolyte Effects of Eplerenone”.
Fast-forward a decade to a study from Israel, by Ori and colleagues [44]. From 39 consecutive patients referred
with ‘low renin hypertension’ they took 24 who screened positive on ARR but had
a plasma aldosterone level below their cut-off (14.5 ng/l); they then compared
them with 24 patients with confirmed primary aldosteronism, of whom 7 had an
APA. The patients were put on spironolactone (n=46) or eplerenone (n=2) for 3
years, and assessed at 1 and 3 years against baseline. At one year antagonist
dose averaged 33 mg/day, and at 3 years 29 mg/day. Despite these modest doses
the authors reported highly significant results. The requirement for other
antihypertensives fell from 2.6 to 1.5, on average; SBP from 149±14 to 126±12;
DBP from 88±10 to 78±7 mm Hg; and left ventricular mass index from 143±25 to
118±20, all at one year. These changes persisted, and for some became more
marked, at 3 years.
What is striking is that absolutely no difference between the 2 groups of 24 was
seen. The authors’ conclusion is that “in patients with PA/low renin
hypertension, long-term regression of left ventricular hypertension may be
achieved with low-dose mineralocorticoid receptor blockers”. True, but in the
opinion of the present author, there are 2 other salient conclusions that can be
drawn from the study. The first is that perhaps ~ 60% (24/39) patients with
so-called low renin hypertension may in fact have occult primary aldosteronism.
Secondly, and the reason why such patients were consigned to the LRH category,
is that for them the 14.5 ng/dl cut-off for aldosterone used in this study was
pitched too high. What this underlines is that whatever the PAC, if the renin
levels are sufficiently low to yield an indicative ARR, the patient should
undergo confirmatory/exclusion testing.
5. ACTH: the silent confounder?
The fifth area addresses a cognate question, but deserves a separate section. In
2011 Gouli et al. [45] published a paper
in which they detailed a further variation of the fludrocortisone saline
suppression test (FST) – the dexamethasone-enhanced fludrocortisone saline
suppression test (FDST). In 80 normotensives with normal adrenals on imaging who
underwent an FDST, the upper limit (97.5%) of normal PAC was set at 3.0 ng/dl,
with 2 mg dexamethasone given at midnight before the final day of study. When
the same test was administered to 130 consecutive hypertensives, 29% showed PAC
values above the upper limit found in control, non-hypertensive subjects. The
interpretation offered of this study is that ‘normal’ levels of PAC are to a
surprising extent driven by ACTH, and that as a consequence, a diagnosis of
autonomous aldosterone secretion, at lower levels, is thus lost in the noise.
The inevitable conclusion from this interpretation is that autonomous
aldosterone secretion is at least in part the driver of≥30% of so-called
essential hypertension.
An additional, potential and mechanistically puzzling role for ACTH in PA is that
of a driver, rather than a confounder, and was very recently published from the
same group [16]. A group of essential
hypertensives – that is, negative on ARR and FDST – were given ultra-low dose
ACTH and their aldosterone and cortisol responses compared with control
normotensive subjects. The hypertensives fell into 2 remarkably tight groups, on
the basis of a briskly elevated PAC in response to ACTH: hyper-responders showed
a marked rise whereas non-responders had PAC levels indistinguishable from
control, and which rose to levels<20% of those in hyper-responders; plasma
cortisol levels rose indistinguishably in all groups. Subjects were then
subjected to a treadmill stress test: the hyper-responders to ultra-low dose
ACTH showed a major increase in PAC, in contrast with a much lesser (~ 20%)
change in controls and non-responders: in all 3 groups cortisol levels were
indistinguishable, and did not rise.
The interpretation of these studies is that hyper-responders (27% of ‘non-PA’
patients on ARR/FDST criteria) have an adrenal glomerulosa which is extremely
sensitive to very small increases in ACTH levels, such as might be seen in
episodic minor stress, and that a concatenation of such episodes might thus
produce a net daily production of aldosterone above normal.
Between them, these studies have the potential to redefine primary aldosteronism.
They include some unusual features – persuading control subjects to undergo
imaging and then a 4-day FDST; remarkably low variations in plasma aldosterone
and cortisol levels in the ultra low-ACTH test; a renin level not low enough to
exclude a proportion of the hyper-responders on the basis of ARR, given that
their mean PAC was 2.8 ng/dl, which means a reasonable number would have
PAC>3 ng/dl, the FDST cut-off, as previously noted [45].
What this means is that these potential roles of ACTH need to be further studied,
not by slavish repetition of the above studies, but by a slightly different
approach. Persuading normal subjects to undergo an FDST is a very tall order:
ideally, then, the baseline PAC values might be set using a
dexamethasone-enhanced seated saline suppression test (DSSST), followed by its
application to a cohort of hypertensives. If the results – as expected – are
similar to those of Gouli et al. [45],
then hypertensives with a PAC below the upper limit of normal might undergo an
ultra-low dose ACTH test, to seek to reproduce the findings in Markou et al.
[16] cited above.
6. Low dose MR antagonist as part of first-line therapy
The sixth, and last, clinical question is the public health issue inherent in
primary aldosteronism. The risk profile of PA is considerably higher than in
age-, sex- and BP-matched essential hypertension [46]. The prevalence of hypertension in
adults is commonly quoted as ~ 40% in Japan, and 29% in the USA. On current
population figures, plus a figure of PA being 10% of hypertension, this means
almost 5 million subjects with PA in Japan, and over 10 million in the USA. In
no country at present are sufficient numbers of hypertensives screened to
identify 1% of patients with PA; in 99+% of such subjects, the condition remains
occult over their lifetime, truly the tip of the iceberg.
In addition, no country has the capacity, or the available resources, to manage
the additional 99+% of patients by the pathways currently recommended [47]. Once an ARR is positive, the patient
is then committed to confirmatory/exclusion testing, imaging, AVS and possibly
surgery. This is a time-consuming, costly exercise; in addition to cost, there
are, for example, very few interventional radiologists able to safely and
reproducibly perform AVS.
The question then is what to do? The answer proffered is to include a low dose
mineralocorticoid receptor antagonist into first-line treatment for all
hypertensives with reasonable renal function. It is safe and efficacious in
truly essential hypertension [38],
selective in resistant hypertension [48],
and game changing in PA. The model is that of vaccination of 12 year old girls
against human papilloma virus to prevent carcinoma of the cervix. We do not
immunise half that population, and then wait 40 years to compare the outcomes.
Primary aldosteronism is far more common in women than cervical cancer, but far
less immediately bleak; it deserves at least a similar degree of intervention to
protect those at risk.
7. New answers: pathophysiology: passé or phoenix?
One final note, to revert from the clinical back to basic science. In the
excitement of molecular genetics, we may be excused for jettisoning
pathophysiology, though at our peril. In the mountains of New Guinea during the
monsoon season the average daily sodium intake is 2–3 meq; in response
aldosterone levels are sky-high, blood pressure low normal and there is
absolutely no cardiovascular damage. For the latter what is needed are
aldosterone levels inappropriate for salt status; we do not know how salt acts
under these circumstances. One possible way is to elicit the protracted
over-secretion of endogenous ouabain-like (EO) molecules, as has been previously
canvassed in full [49]. EO rises, like
aldosterone in response to ACTH, angiotensin II (but via AT2R) and –
in sharp contrast to aldosterone – to sodium loading. EO acts on
membrane-located Na+/K+ATPase to cause vasoconstriction
and resultant tissue damage. If this is the case, then the deleterious
cardiovascular effects of autonomous aldosterone secretion may not primarily lie
in its (unlikely) ability to be directly cardiotoxic – for example, in
cardiomyocytes unprotected by 11βHSD2 – but its ability to over-reabsorb sodium
in the renal tubules, and to raise EO levels as the proximate cause of
cardiovascular damage.
If this were shown to be the case it would bring considerable satisfaction not
just to nephrologists, who have always stoutly maintained the primacy of the
kidney, but also to clinicians and physiologists. For clinicians, it would
validate using ENaC inhibitors (amiloride, triamterine) routinely in conjunction
with low dose mineralocorticoid receptor antagonists; for physiologists, it
would mean the answer to a long-vexing conundrum, in demonstrating the
pathophysiologic role of a pathway which presumably evolved to produce a
pressure diuresis after sodium loading, when the requisite lowering of
aldosterone levels cannot be accomplished.
If this conundrum were answered over the next 5 years it would represent a
triumph for interactive medicine, systems biology – or whatever is currently
modish name for physiology.
