Autism: a spectrum, dimensions or clusters? How about a multi-dimensional cluster of spectrums?

A write-up (see here) of the paper by Hyunsik Kim and colleagues [1] was the initial impetus for formulating this blog post, but it quickly escalated into something a little larger when the findings from Frank Duffy & Heidelise Als [2] also popped up.

The question at hand: how should one conceptualise autism? Is it truly a spectrum as per the Lorna Wing proposition, or is it something a trifle more complicated? As per the title of this post, should we perhaps be thinking about autism as some sort of "multi-dimensional cluster of spectrums?" I'll come back to that idea shortly.

Well, it's not for me to make definitive conclusions on this blog. Science rarely, if at all, provides an absolute 'truth' but rather the probability that something is approaching truth. Such a notion goes double when you consider the singular label of autism and the huge heterogeneity that it encompasses. There are no easy answers and probably little or no truths.

Starting with the Kim paper (including some notable names such as the surname 'Gadow') and the name of the research game was modelling, modelling in a computational sense. So: "The sample comprised 3,825 youth, who were consecutive referrals to a university developmental disabilities or child psychiatric outpatient clinic." The CASI-4R - formulated by Prof. Gadow - was the schedule administered, which includes "an ASD [autism spectrum disorder] symptom rating scale" among other things. Some nifty statistics were applied to the data and the initial findings were 'tested' on a further group of over 2500 children.

Results: "Based on comparison of 44 different models, results indicated that the ASD symptom phenotype is best conceptualized as multi-dimensional versus a categorical or categorical-dimensional hybrid construct." And the dimensions mentioned in that 'multi-dimensional' statement? Well, lucky for us they were something familiar: "social interaction, communication, and repetitive behaving."

Then to the Duffy/Als paper (again, these authors are no stranger to autism research) and a similar starting point: "The authors postulate that the broad definition of an omnibus 'spectrum disorder' may inhibit delineation of meaningful clinical correlations." Indeed, very familiar (see here). The conclusion: "evidence that an objectively defined, EEG [electroencephalogram] based brain measure may be helpful in illuminating the autism spectrum versus subgroups (clusters) question." The tool used by Duffy/Als in their study was something called NbClust "specifically designed to provide an objective means, i.e. independent of investigator choice, to identify the ‘optimal’ cluster number within a population." Said tool was applied to EEG data derived from 400 participants diagnosed with an ASD. Statistics and more statistics applied to the data revealed that: "430 subjects diagnosed as being on the autism “spectrum” and represented by 40 EEG coherence factors..., fell into two distinct clusters." These autism spectrum clusters differed from each other and importantly, from "554 subject neuro-typical control group subjects, not involved in the clustering process." Interesting results but an unfortunate use of the term 'neurotypical' (see here). Duffy & Als conclude that their data support a view whereby "autism disorder should not be seen as a continuous spectrum." So Kim & Duffy/Als arrive at similar conclusions: a singular 'spectrum' idea of autism is probably not the best way of conceptualising the essence of the label.

I would perhaps add in a little more evidence for the idea that 'multi-dimensional clusters of spectrums' is a potentially better fit. I used the words 'spectrums' (plural) because there is a growing body of evidence to support the idea of more than one 'type' of autism. I say that from the perspective of evidence for autism being 'acquired' under several different circumstances (e.g. accompanying inborn errors of metabolism, linked to exposure to certain infections or diseases, etc). There's also evidence that clinical profiles under the umbrella term autism are not uniform (e.g. regressive autism, the so-called 'optimal outcomers', differing developmental trajectories, etc). And when one looks at something like the success (or not) of intervention, it's plain to see that there is no universally shared genetics and/or biology of autism in the singular either (see here and see here for examples). Add in the idea that autism rarely appears in a diagnostic vacuum (see here) and that said comorbidity might 'cluster' in some subgroups of autism (see here), and I hope you can see why 'plural' might be a good addition to any attempt to re-conceptualise autism: spectrum, dimension, tapestry, cluster or however you think it should be defined...

Oh, and since we're on the topic of trying to conceptualise autism, a new book out recently has been reviewed in Nature (see here). It talks about how "conclusive findings about sex-linked brain differences have failed to materialize" which is particularly apt in relation to previous talk about 'extreme male brains' as a way of conceptualising [some] autism (see here). One quote I particularly liked from the review is this one: "The brain is no more gendered than the liver or kidneys or heart."

----------

[1] Kim H. et al. Quantifying the Optimal Structure of the Autism Phenotype: A Comprehensive Comparison of Dimensional, Categorical, and Hybrid Models. J Am Acad Child Adolesc Psychiatry. 2018 Oct 29. pii: S0890-8567(18)31894-X.

[2] Duffy FH. & Als H. Autism, spectrum or clusters? An EEG coherence study. BMC Neurol. 2019 Feb 14;19(1):27.

----------

More welcome research on post-exertional malaise (PEM) in ME/CFS

The findings reported by Lily Chu and colleagues [1] provide the blogging fodder today, and some further welcome research into the concept of post-extertional malaise (PEM) in the context of chronic fatigue syndrome / myalgic encephalomyelitis (CFS/ME).

PEM represents a cardinal feature of CFS/ME, where physical exertion - whether covering exercise or even just daily activities - brings about a period of (even more) exhaustion, weakness and fatigue. As per my previous discussion of the paper by McManimen and colleagues [2], there's still some ambiguity when it comes to clinically defining PEM (and indeed, whether the word 'malaise' covers the range of symptoms that the term includes). What is clear however is that PEM is real and can be absolutely disabling.

The Chu study, including the name Jose Montoya on the authorship list, set out to "describe symptoms associated with and the time course of PEM." Authors designed an online survey "asking subjects about the history and course of their ME/CFS" that asked a few questions about PEM. This included: "1) What symptoms, if any, are triggered or worsened by physical or mental activity?; 2) What symptoms are triggered or worsened by emotional distress?; 3) How soon usually after starting mental or physical exertion does your illness begin to worsen?; and 4) If you feel worse after activities, how long does this worsening usually last?" They also describe how they were careful not to actually use the word 'PEM' during the course of their research in order to "try to decrease the chances that subjects already diagnosed with ME/CFS would automatically answer our question based on their knowledge of or preconceived notions about PEM." Smart move.

Results: some 150 people as part of something called the MGEISD (Genetic Expression and Immune System Dynamics) study responded to the survey. While some of the participants were formally diagnosed with ME/CFS, the cohort also included those "waiting to be seen at the clinic, members of online ME/CFS forums, and participants of local ME/CFS support groups." Although all were telephone screened to see "if they fitted Fukuda 1994 CFS criteria" one does need to be a little careful with describing such a process as providing anything approaching a 'homogeneous' group.

"Most subjects (N = 129, 90%) experienced PEM with both physical/ cognitive exertion and emotional distress." There didn't seem to be any specific 'rules' in terms of the connection between the 'stressor' and onset of PEM, although around 40% of participants reported experiencing PEM within 24 hours. This was however subject to quite some variation. Symptoms reported in relation to the PEM questionnaire items were also variable. Fatigue is up there as one would expect, but also things like sleeping issues and headache were noted. Around 40% of participants also reported on a constellation of PEM symptoms: fatigue, sleep disturbance, pain and "at least one immune-related symptom."

What is becoming a little clearer from this and other research is that (a) PEM or whatever you want to call it is a real issue for many people diagnosed with CFS/ME and (b) the symptoms of PEM are not necessarily uniform in either character or timescale (onset, duration, etc.) What's still missing? Well, biology is still missing from quite a lot of the PEM literature; biology in terms of what PEM looks like on a physiological level and whether there may be something that can be done to alleviate it at a biological level aside from resting up (often for days at a time). I've talked about this before on this blog, in terms of biologically characterising PEM and how, for example, it might look in terms of defining recovery from ME/CFS (see here). To that end, lots more investigation in this area is implied...

And while on the topic of ME/CFS, I'll draw your attention to the paper by Karfakis [3] talking about the "biopolitics of CFS/ME." Yes, there has been and continues to be lots of that, including the phrase: "CFS/ME is an illness trapped between medicine and psychology". Discuss in 500 words.

----------

[1] Chu L. et al. Deconstructing post-exertional malaise in myalgic encephalomyelitis/ chronic fatigue syndrome: A patient-centered, cross-sectional survey. PLoS One. 2018 Jun 1;13(6):e0197811.

[2] McManimen SL. et al. Deconstructing post-exertional malaise: An exploratory factor analysis. Journal of health psychology. August 2016:1359105316664139.

[3] Karfakis N. The biopolitics of CFS/ME. Stud Hist Philos Biol Biomed Sci. 2018 Jun 8. pii: S1369-8486(17)30070-5.

----------

What does 'recovery from myalgic encephalomyelitis / chronic fatigue syndrome' look like?

The title of this post reflects the material included in the paper by Andrew Devendorf and colleagues [1] who sought "an operationalised definition of recovery from myalgic encephalomyelitis (ME) and chronic fatigue syndrome (CFS) for research and practice."

Although perhaps at first sight being a rather simple question to answer - Q: What does recovery look like in relation to ME/CFS? A: Complete remission of symptoms -  there's a lot more behind such a query than you might expect. I'm particularly thinking about how 'recovery' has been a real source of discussion/debate/argument (delete as appropriate) in the context of ME/CFS in recent 'PACE' times (see here and see here).

Ten experts on ME/CFS were quizzed about their views specifically on recovery from the condition(s). As probably expected: "Physicians conceptualised recovery as complete symptom remission and a return to premorbid functioning (adjusted for with age)." No surprises there then. Insofar as the term 'significant improvement', experts also quite sensibly reported that this should be 'operationalised' as "a substantial reduction in symptoms with considerable functional gains, where patients may operate in daily life but still must cope or be treated." Again, not really that earth-shattering to be honest.

'Recovery' in the context of ME/CFS has been discussed in the peer-reviewed domain before [2] with views to "recommend a consistent definition that captures a broad-based return to health with assessments of both fatigue and function as well as the patient's perceptions of his/her recovery status." Although not everyone agreed with all that was said by Adamowicz and colleagues [3], there  is a general consensus that recovery represents an absolute term where 'complete symptom remission' is the important feature. Everything else outside of recovery falls into 'a spectrum of improvement' (substantial, significant, so-so, etc) through to no change or even potential symptom worsening.

The task facing ME/CFS research and practice now: how best to measure the recovery / non-recovery spectrum? I agree that questionnaires about fatigue and (very) important clinical signs like post-exertional malaise (PEM) need to be part and parcel of such measurements. As with any condition that is [currently] diagnosed solely on the basis of presented symptoms with no genetic or biological test yet able to distinguish cases from not-cases, questionnaires and interviews are always going to be important in terms of how symptoms change/evolve over time either naturally or as a result of some specific intervention(s). But research and practice really need more than that; they need some objectivity too...

So, physical activity levels: well, I've often gone on (and on!) about the great potential of actigraphy in the context of several research areas. You want data on rest and activity cycles for a condition characterised by rest and activity cycles? There's plenty of technology out there in the marketplace to measure such cycles and associated measures that could be used in a complementary fashion (see here). How do you define 'recovery' based on such data? Well, typically (at the moment where such tech are still relatively new) there probably won't be 'before and after' data as noted in other examples of such tech use (see here) so a reliance of general population data is the next best thing. Yes, by doing so we're moving from individual data to more generalised data as a comparator, but when such generalised data runs into the thousands or even millions of people (see here for one example) you can start to build up a picture based on 'expected' physical activity levels taking into account sex/gender, age and other important variables. The important point is that one element of recovery in CFS/ME is going to be a restoration of physical activity levels (and perhaps even sleep cycles?) and there are easy ways to measure that minus any subjectivity issues potentially included in questionnaires or interviews. I say all of that acknowledging that levels of typical day-to-day tasks such as work, shopping, socialising and other pastimes/activities which generally require physical exertion can also be to some extent ascertained via a questionnaire.

PEM? I'm not so sure about how to measure this outside of some limited research in this area [4] talking about a "change in complement C4a level and the increase in pain and fatigue 24 h following the self-paced, physiologically limited exercise support the use of C4a as a marker for postexertional malaise in people with ME⁄CFS." Complement C4a by the way, is something of a 'mediator of local inflammatory processes' and has cropped up on this blog before (see here). In light of other work also suggesting that PEM might have cognitive effects [5] too, I'm sure a little more scientific inquiry could bring together a suite of fairly cost-effective ways and means to objectively measure PEM both in terms of biology and also cognitive prowess. Yes, there are details to be worked out but...

Other things to be included in recovery criteria? Well, it depends on whether you're going to go down the sole 'core' symptoms pathway or going to include elements that may potentially be more peripheral [6] yet still impact on notions of recovery in CFS/ME (hyperacuity anyone?). That's not my call I'm afraid, although given what has been discussed with regards to health-related quality of life and ME/CFS (see here), the addition of something like the EQ-5D-3L to any protocol might not also go amiss as part of any measurement of recovery with CFS/ME in mind...

----------

[1] Devendorf AR. et al. Defining and measuring recovery from myalgic encephalomyelitis and chronic fatigue syndrome: the physician perspective. Disabil Rehabil. 2017 Oct 5:1-8.

[2] Adamowicz JL. et al. Defining recovery in chronic fatigue syndrome: a critical review. Qual Life Res. 2014 Nov;23(9):2407-16.

[3] Twisk FN. A definition of recovery in myalgic encephalomyelitis and chronic fatigue syndrome should be based upon objective measures. Qual Life Res. 2014 Nov;23(9):2417-8.

[4] Nijs J. et al. Unravelling the nature of postexertional malaise in myalgic encephalomyelitis/chronic fatigue syndrome: the role of elastase, complement C4a and interleukin-1beta. J Intern Med. 2010 Apr;267(4):418-35.

[5] Cook DB. et al. Neural consequences of post-exertion malaise in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome. Brain, Behavior, and Immunity. 2017; 62: 87-99.

[6] Whiteley P.  et al. Correlates of Overlapping Fatigue Syndromes. Journal of Nutritional & Environmental Medicine. 2004; 14: 247-259.

----------

Post-exertional malaise (PEM) in CFS might mean more than one thing

"The results suggest that post-exertional malaise [PEM] is composed of two empirically different experiences, one for generalized fatigue and one for muscle-specific fatigue."

So said the findings reported by Stephanie McManimen and colleagues [1] looking at one of the most common and debilitating aspects of chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME) and how rough-and-ready generalised descriptions often do little to reveal the complexities of this particular symptom.

The research group at the centre of this new paper based at DePaul University under the stewardship of Leonard Jason have some previous interest in this facet of ME/CFS as per other recent publications [2] talking about the trials and tribulations of defining PEM. Once again, the very multiple and very complicated ways that CFS/ME is currently diagnosed (see here for example) comes into play as the wording used to define PEM seems to count in terms of who is more or less likely to display this symptom. I might also add that even the words 'post-exertional malaise' have been subject to question and redefinition as per the use of the term PENE (Postexertional neuroimmune exhaustion) [3].

In their latest paper, McManimen et al set out to "discern whether post-exertional malaise is a unified construct or whether it is composed of two smaller constructs, muscle fatigue and generalized fatigue." As per the opening sentence of this post, researchers came down on the side of PEM meaning more than one thing, and with it the possibility of further revisions being required to the criteria to define this concept. Indeed, going back to yet another paper from this group [4] researchers suggested that several composite items might better define PEM over just one question: "Dead, heavy feeling that occurs quickly after starting to exercise; Next day soreness or fatigue after non-strenuous, everyday activities; Mentally tired after the slightest effort; Physically drained or sick after mild activity; and Minimum exercise makes you physically tired." Words, in a diagnostic sense, need to be used accurately.

And finally, whilst on the topic of CFS/ME, there have been some further developments around the science of CBT and graded exercise...

----------

[1] McManimen SL. et al. Deconstructing post-exertional malaise: An exploratory factor analysis. J Health Psychol. 2016 Aug 24. pii: 1359105316664139.

[2] Jason LA.  et al. Problems in Defining Post-Exertional Malaise. Journal of prevention & intervention in the community. 2015;43(1):20-31.

[3] Carruthers BM. et al. Myalgic encephalomyelitis: International Consensus Criteria. Journal of Internal Medicine. 2011;270(4):327-338.

[4] Jason LA. et al. Fatigue Scales and Chronic Fatigue Syndrome: Issues of Sensitivity and Specificity. Disabil Stud Q. 2011 Winter;31(1). pii: 1375.

----------

McManimen SL, Sunnquist ML, & Jason LA (2016). Deconstructing post-exertional malaise: An exploratory factor analysis. Journal of health psychology PMID: 27557649

On autism spectrum disorder [research] validity

Today I'm directing your reading attention to a really, really interesting paper by Lynn Waterhouse and colleagues [1] (open-access) whose review findings suggest that: "the ASD [autism spectrum disorder] diagnosis lacks biological and construct validity."

The paper is a bit of a long read but most definitely worth it as the quite complicated subject of exactly what goal the label of autism actually serves is discussed. The results of various questions posed by the authors suggest: "No unitary ASD brain impairment or replicated unitary model of ASD brain impairment exists. ASD core diagnostic symptoms are not uniquely linked and are only very rarely expressed without nondiagnostic symptoms. ASD has no reliable early predictor, no unitary developmental course, no unitary life outcome, no unitary recurrence risk, no unitary pattern of BAP [broader autism phenotype] features, and no standard homogeneous subgroups." They conclude that from a research perspective at least, disbanding the label of autism as it currently stands is the next logical step. Said disbanding "is likely to be reductive and uncomfortable" particularly when it comes to all those grand [sweeping] theories of autism put forward down the years. Feathers would not doubt be ruffled.

The authors do make reference to two important concepts when it comes how we might want to rethink autism: the Research Domain Criteria framework (RDoC) and Early Symptomatic Syndromes Eliciting Neurodevelopmental Clinical Examinations (ESSENCE). One is an attempt to move away from simple psychiatric labels as somehow denoting homogeneity, the other is the recognition that labels rarely appear in some sort of diagnostic vacuum. Both are in some way the future of autism research and indeed, the future is already now.

I'm impressed with the paper from Waterhouse et al. The authors have done a good job of basically saying that as things stand, one single label covering such a diverse and heterogeneous group is not fit for purpose. To see real progress in autism research, science needs to think more about those 'autisms' (see here) and stop using the label of autism as the starting point for research (see here). I struggle to disagree with both those sentiments and other authors appear to have reached similar conclusions [2]. Exactly what that means for the autism in the future - from both a research and clinical perspective - is still a little up in the air but the label has weathered change before and no doubt will continue to do so.

And if that isn't enough reading material for you, how about the latest instalment from the British Psychological Society here in Blighty when it comes to autism? Perhaps this will need a revision or two as the Waterhouse suggestions start to percolate through the research community?

----------

[1] Waterhouse L. et al. ASD validity. Review Journal of Autism and Developmental Disorders. 2016. Aug 10.

[2] Geier DA. et al. Examining genotypic variation in autism spectrum disorder and its relationship to parental age and phenotype. Appl Clin Genet. 2016 Jul 28;9:121-9.

----------

Waterhouse, L., London, E., & Gillberg, C. (2016). ASD Validity Review Journal of Autism and Developmental Disorders DOI: 10.1007/s40489-016-0085-x

On defining chronic fatigue syndrome / myalgic encephalomyelitis (CFS/ME)

Most people who know a little bit about chronic fatigue syndrome / myalgic encephalomyelitis (CFS/ME) will probably understand the potential importance of the findings reported by Leonard Jason and colleagues [1] (open-access available here). Suggesting that there may be "four groupings of patients" when it comes to how we "name and define the illnesses", this research group who surveyed over 500 people "in the United States, Great Britain, and Norway" report on one of the biggest challenges facing ME/CFS... how do we accurately define it?

The problem of defining CFS/ME stems from the fact that there are currently several clinical presentations where CFS/ME might figure (or at least where patients fulfil ME/CFS criteria) and several different ways of clinically defining the condition(s) (see here and see here). That list of definitions may indeed also be growing (see here). With all that confusion about clinical overlap and what criteria are defining what patient group, it's little wonder that research is just starting to come to grips with some of the potential underlying biology of [some] CFS/ME (see here) after so many years in the scientific wilderness. Indeed, as with many conditions resting in that 'unexplained symptoms' category, not knowing can sometimes be fertile ground for various [unfounded] theorising...

Jason et al (whose research has graced this blog before) set about categorising their participant group based on various case definitions and symptom presentations. They reported four groupings; by far the largest group of their participants (n=346) fell into a categorisation that "involves more specific criteria" defining CFS/ME including use of the relative new term SEID (Systemic Exertion Intolerance Disease). Further: "efforts have specified domains of substantial reductions of activity, post-exertional malaise, neurocognitive impairment, and sleep dysfunction" and "Patients with these characteristics were more functionally impaired than those meeting just chronic fatigue criteria." The term used by authors for this group was Neuroendocrineimmune Dysfunction Syndrome (NDS) following on from previous work [2]. Smaller numbers of people included on their study did meet criteria for chronic fatigue (greater than 6 months) with (n=47) or without (n=62) explanation "by medical or psychiatric conditions." Jason and colleagues also defined a smaller group who met the Ramsay ME criteria and who were described as an "even more impaired group." I wonder if this might include the 'housebound' group described in other research?

The authors accept that there are 'limitations' to their research including issues around how they chose their sample(s) and some of the tools used relying on self-report; further work is indicated. But as a start trying to disentangle both the heterogeneity and issues with clinical classification when it comes to CFS/ME, their paper represents a good attempt to further focus minds on the spectrum of various overlapping fatigue conditions present and how we go about teasing apart phenotypes with a focus on core symptoms [2]. Minus the psychiatry focus, perhaps it is time to also looking at applying something like the principles of RDoC to the issue of ME/CFS?

----------

[1] Jason LA. et al. Case definitions integrating empiric and consensus perspectives. Fatigue. 2016;4(1):1-23.

[2] Jason LA. et al. Factor Analysis of the DePaul Symptom Questionnaire: Identifying Core Domains. J Neurol Neurobiol. 2015 Sep;1(4).

----------

Jason LA, McManimen S, Sunnquist M, Brown A, Furst J, Newton JL, & Strand EB (2016). Case definitions integrating empiric and consensus perspectives. Fatigue : biomedicine, health & behavior, 4 (1), 1-23 PMID: 27088059

Just what is 'non-coeliac gluten sensitivity'?

The question posed in the title of this post relates to the findings reported by Federica Branchi and colleagues [1] (open-access available here) surveying "knowledge about nonceliac gluten sensitivity among gastroenterology specialists" in Italy.

In amongst the various results gathered from over 200 gastroenterologists who completed a study questionnaire, the authors report that: "98.6% were aware of the existence of a syndrome called "nonceliac gluten sensitivity" and 77% believe in its existence." Further however: "44% of the specialists are not able to give its correct definition" defined by the authors as: "a heterogeneous disorder induced by gluten ingestion in which neither allergic nor autoimmune mechanisms seem to be involved."

These are interesting results bearing in mind the very simplistic questionnaire used, that "consisted of 12 multiple-choice questions inquiring about NCGS (nonceliac gluten sensitivity)" and the pretty poor response rate (212/1322; 16%). I would tend to agree with the authors that there is quite a bit more to do "to improve the knowledge about these disorders, as well as promote the circulation of unambiguous and scientifically relevant information."

NCGS is something that has been of real interest to me for quite a while (see here). Distinct from more traditional conditions linked to gluten consumption (such as the autoimmune condition coeliac disease) I am fascinated by the idea that there may be a spectrum of gluten-related ills to pay attention to (see here). For those that have followed the NCGS journey, you'll probably already know that the research road travelled has been rocky at times (see here) as 'not knowing' can sometimes open up real divisions. That being said, there is some important 'controlled' evidence to suggest that there is something to see when it comes to NCGS (see here) also potentially spilling over into non-gastroenterological manifestations too (see here). I might add that diagnoses such as autism are also probably not protective of conditions appearing under the NCGS banner either (see here).

And if you want some of the current best data on a consensus for NCGS, look no further than here [2] and here [3]...

----------

[1] Branchi F. et al. Management of Nonceliac Gluten Sensitivity by Gastroenterology Specialists: Data from an Italian Survey. Biomed Res Int. 2015;2015:530136.

[2] Catassi C. et al. Non-Celiac Gluten sensitivity: the new frontier of gluten related disorders. Nutrients. 2013 Sep 26;5(10):3839-53.

[3] Lebwohl B. et al. Celiac disease and non-celiac gluten sensitivity. BMJ. 2015 Oct 5;351:h4347.

----------

Branchi F, Ferretti F, Norsa L, Roncoroni L, Conte D, Bardella MT, & Elli L (2015). Management of Nonceliac Gluten Sensitivity by Gastroenterology Specialists: Data from an Italian Survey. BioMed research international, 2015 PMID: 26665005