Least Popular Posts of 2017

2017 was a really bad year. The U.S. is more divided than ever, the truth is meaningless, well-researched journalism is called FAKE NEWS, the President lies once every minute, white supremacist rallies have been normalized, some tech companies¹ continue to invade our privacy/extract personal data, exploit the middle and lower classes,² and displace long-time residents from urban areas. And who knows what health care and Alaska will look like in 2018.

Yes, this is classic Neurocritic pessimism.³

While everyone else rings in the New Year by commemorating the best and brightest of 2017 in formulaic Top Whatever lists, The Neurocritic has decided to wallow in shame. To mark this Celebration of Failure, I have compiled a Bottom Five list,⁴ the year's least popular posts as measured by Google Analytics. The last time I compiled a “Worst of” list was in 2012.

Methods: The number of pageviews per post was copied and pasted into an Excel file, sorted by date. Then the total pageviews for each post was prorated by the vintage of the post, to give an estimate of daily views.⁵ 

Results: The posts are listed in inverse order, starting with #5 and ending with #1 (least popular).

5 Most Unpopular Posts of 2017

5. Terrorism and the Implicit Association Test – I actually worked pretty hard on this one. It's about the stereotyping of Muslims, the importance of language (e.g., Theresa May: “the single, evil ideology of Islamist extremism that preaches hatred, sows division, and promotes sectarianism”), a demonstration that semantics derived automatically from language corpora contain human-like biases, the Arab-Muslim IAT (which found little to no bias against Muslims), and some general problems with the IAT.

4. Smell as a Weapon, and Odor as Entertainment – This was from my two-part olfactory series, which covered the interesting history of Olfactory Warfare (e.g, stink bombs, stealth camouflage) and the use of smell in cinematic and VR contexts. {or at least, it was interesting to me}.

3. The Big Bad Brain – This featured a fun and catchy music video (High) by Sir Sly, which was an earworm for me. But too esoteric and not much staying power.

2. What's Popular at #CNS2017? – This falls under the perennially unpopular category of “yearly conference announcements”, which is only relevant around the time of the meeting.

1. Olfactory Deterrence – This was about the prospect of nuclear war and how putrid smells might deter the use of nuclear weapons, along with eradicating cavalier attitudes about them.


Discussion: We can easily see some themes emerging: the IAT, olfaction, music videos, and the Cognitive Neuroscience Society meeting.

Conclusion: People are sick of the IAT, aren't thrilled about the sense of smell (especially in relation to nuclear war), and do not like music videos or CNS Meeting announcements. However, they do like meeting recaps, as shown by the popularity of What are the Big Ideas in Cognitive Neuroscience? and The Big Ideas in Cognitive Neuroscience, Explained.


Footnotes

¹ Uber deserves special mention.

² This one is from 2016, but it's a real eye-opener: The Not-So-Wholesome Reality Behind The Making of Your Meal Kit.

³ This has been the worst-ever year for me personally as well, so I see no reason to be optimistic.

⁴ Actually, #5 is Survival and Grief. I cannot bear to feature this one, so the closely ranked #6 is a stand-in.

⁵ The post with the absolute lowest number of views (Brief Guide to the CTE Brains in the News. Part 2: Fred McNeill) was written on 12/11/2017. For a true reading of yearly “staying power” we'd need to follow all posts for 365 days.

Amygdala Stimulation in the Absence of Emotional Experience Enhances Memory for Neutral Objects

The amygdala is a small structure located within the medial temporal lobes (MTL), consisting of a discrete set of nuclei. It has a reputation as the “fear center” or “emotion center” of the brain, although it performs multiple functions. One well-known activity of the amygdala, via its connections with other MTL areas, involves an enhancement of memories that are emotional in nature (compared to neutral). Humans and rodents with damaged or inactivated amygdalae fail to show this emotion-related enhancement, although memory for neutral items is relatively preserved (Adolphs et al., 1997; Phelps & Anderson, 1997; McGaugh, 2013).

A new brain stimulation study (Inman et al., 2017) raises interesting questions about the necessity of subjective emotional experience in the memory enhancement effect. A group of 14 refractory epilepsy patients underwent surgery to implant electrodes in the left or right amygdala (and elsewhere) for the sole purpose of monitoring the source of their seizures. In a boon for affiliated research programs everywhere, patients are able to participate in experiments while waiting around for seizures to occur.

The stimulating electrodes were located in or near the basolateral complex of the amygdala (BLA), shown below. The stimulation protocol was developed from similar studies in rats, which demonstrated that direct electrical stimulation of BLA can improve memory for non-emotional events when tested on subsequent days (Bass et al., 2012; 2014; 2015).

Fig. 1A and B (modified from Inman et al., 2017). 
(A) A representative postoperative coronal MRI showing electrode contacts in the amygdala (white square). (B) Illustration of left amygdala with black circles indicating estimated centroids of bipolar stimulation in or near the BLA in all 14 patients. White borders denote right-sided stimulation.


The direct translation from animals to humans is a clear strength of the paper (Inman et al., 2017):

...direct activation of the BLA modulated neuronal activity and markers of synaptic plasticity in the hippocampus and perirhinal cortex, two structures important for declarative memory that are directly innervated by the BLA.  ... These and other studies [in animals] have led to the view that an emotional experience engages the amygdala, which in turn enhances memory for that experience through modulation of synaptic plasticity-related processes underlying memory consolidation in other brain regions. This model predicts that direct stimulation of the human amygdala could enhance memory in a manner analogous to emotion’s enhancing effects on long-term memory.

The experimental task was a test of object recognition memory. Pictures of 160 neutral objects were presented on Day 1 while the participants made “indoor” or “outdoor” decisions (which were quite ambiguous in many cases). The purpose of this task was to engage a deep level of semantic encoding of each object, which was presented for 3 seconds. Immediately after stimulus offset for half the items (n=80), a train of electrical stimulation pulses was presented for 1 second (each pulse = 500 μs biphasic square wave; pulse frequency = 50 Hz; train frequency = 8 Hz). For the other half (n=80), no stimulation was presented. Each trial was separated by a 5 second interval.

Fig. 1D (modified from Inman et al., 2017).


An immediate recognition memory test was presented after completion of the study phase. Yes/no decisions were made on 40 old objects with post-stimulation, 40 old objects with no stimulation, and 40 new objects (“foils”). Then 24 hours later, a similar yes/no recognition test was presented, but this time with the other set of items not tested previously, along with a new set of foils. The prediction was that electrical stimulation of the amygdala would act as an artificial “boost” of performance on the 24 hour test, after memory consolidation had occurred.

This prediction was (mostly) supported as shown below, with one caveat I'll explain shortly. In Panel A, a commonly used measure of discrimination performance (d′) is shown for the Immediate and One-Day tests, with red dots indicating stimulation and blue dots no stimulation (one dot per patient). Most participants performed better on stimulated items regardless of whether on the Immediate test or One-Day test, although variability was higher on the Immediate test. Panel B shows a summary of the performance difference for stimulation − no stimulation trials. Paired-samples t-tests (two sided) were conducted for each recognition-memory interval. The result for One-Day was significant (p=.003), but the result for Immediate was not (p=.30). This would seem to be convincing evidence that amygdala stimulation during encoding enhanced delayed recognition memory selectively.

Fig. 2A and B (modified from Inman et al., 2017).


HOWEVER, from the statistics presented thus far, we don't know whether the memory enhancement effect was statistically larger for the One-Day test. My guess is not, because an ANOVA showed a main effect of test day (p< 0.001) and a main effect of stimulation (p= 0.03). But no interaction between these variables was reported.

Nonetheless, the study was fascinating because the patients were unable to say whether or not stimulation was delivered in a subsequent test of awareness (10 trials of each condition):

All 14 patients denied subjective awareness of the amygdala stimulation on every trial. In addition, no patient reported emotional responses associated with amygdala stimulation during the stimulation awareness test or during recognition-memory testing. Moreover, similar amygdala-stimulation parameters caused no detectable autonomic changes in patients (n = 7) undergoing stimulation parameter screening.

The take-home message is that subjective and objective indicators of emotion were not necessary for amygdala stimulation during encoding to enhance subsequent recognition of neutral material. “This memory enhancement was accompanied by neuronal oscillations during retrieval that reflected increased interactions between the amygdala, hippocampus, and perirhinal cortex”¹ (as had been shown previously in animals).²

So it seems that subjective emotional experience may be an unnecessary epiphenomenon for the boosting effect of emotion in the formation of declarative memories. Or at least in this limited (albeit impressive) laboratory setting. And here I will step aside from being overly critical. Anyone who wants to slam the reproducibility of an n=14 rare patient sample size should be prepared to run the same study with 42 individuals with amygdala depth electrodes.


Footnotes

¹ Inman et al., 2017:

For [n = 5 patients] with electrodes localized concurrently in the amygdala, hippocampus, and perirhinal cortex), local field potentials (LFPs) from each region were recorded simultaneously during the immediate and one-day recognition-memory tests... LFP oscillations were apparent in the theta (here 5–7 Hz) and gamma (30–55 Hz) ranges...  ...  Recognition during the one-day test but not during the immediate test exhibited increased power in perirhinal cortex in the gamma frequency range for remembered objects previously followed by stimulation compared with remembered objects without stimulation. Furthermore, LFPs during the one-day test, but not during the immediate test, revealed increased coherence of hippocampal–perirhinal oscillations in the theta frequency range for remembered objects previously followed by stimulation compared with remembered objects without stimulation.

² If you think the 14 patients with epilepsy were variable, wait until you see the [overly honest] results from even smaller studies with rats.

Fig. S7 (Inman et al., 2017).

Conveniently, Professor Dorothy Bishop has a new blog post on Using simulations to understand the importance of sample size. So yes, sample size matters...


References

Adolphs R, Cahill L, Schul R, Babinsky R. (1997). Impaired declarative memory for emotional material following bilateral amygdala damage in humans. Learn Mem. 4(3):291-300.

Bass DI, Manns JR. (2015). Memory-enhancing amygdala stimulation elicits gamma synchrony in the hippocampus. Behav Neurosci. 129(3):244-56.

Bass DI, Nizam ZG, Partain KN, Wang A, Manns JR. (2014). Amygdala-mediated enhancement of memory for specific events depends on the hippocampus. Neurobiol Learn Mem. 107:37-41.

Bass DI, Partain KN, Manns JR. (2012). Event-specific enhancement of memory via brief electrical stimulation to the basolateral complex of the amygdala in rats. Behav Neurosci. 126(1):204-8.

Ikegaya Y, Saito H, Abe K. (1996). The basomedial and basolateral amygdaloid nuclei contribute to the induction of long-term potentiation in the dentate gyrus in vivo. Eur J Neurosci. 8(9):1833-9.

Inman CS, Manns JR, Bijanki KR, Bass DI, Hamann S, Drane DL, Fasano RE, Kovach CK, Gross RE, Willie JT. (2017). Direct electrical stimulation of the amygdala enhances declarative memory in humans. Proc Natl Acad Sci.  Dec 18. [Epub ahead of print]

McGaugh JL.(2013). Making lasting memories: remembering the significant. Proc Natl Acad Sci 110 Suppl 2:10402-7.

Phelps EA, Anderson AK. (1997). Emotional memory: what does the amygdala do? Curr Biol. 7(5):R311-4.

Brief Guide to the CTE Brains in the News. Part 2: Fred McNeill

Chronic traumatic encephalopathy (CTE) is the neurodegenerative disease of the moment, made famous by the violent and untimely deaths of many retired professional athletes. Repeated blows to the head sustained in contact sports such as boxing and American football can result in abnormal accumulations of tau protein (usually many years later). The autopsied brains from two of these individuals are shown below.

Left: courtesy of Dr. Ann McKee in NYT.  Right: courtesy of Dr. Bennett Omalu in CNN. These are coronal sections¹ from the autopsied brains of: (L) Aaron Hernandez, aged 27; and (R) Fred McNeill, aged 63.


Part 1 of this series looked at complicating factors in the life of Aaron Hernandez — PCP abuse, death by asphyxiation — that presumably had some impact on his brain beyond the effects of concussions in football.

Part 2 will discuss the tragic case of Fred McNeill, former star linebacker for the Minnesota Vikings. He died in 2015 from complications of Amyotrophic Lateral Sclerosis (ALS), suggesting that his was not a “pure” case of CTE, either.


Fred McNeill

McNeill in 1974 (Mike Zerby / Minneapolis Star Tribune).

Obituary: Standout of the 1970s and 1980s was suffering from dementia and died from complications from ALS, according to Matt Blair [close friend and former teammate]

ALS is a motor neuron disease that causes progressive wasting and death of neurons that control voluntary muscles of the limbs and ultimately the muscles that control breathing and swallowing. Around 30-50% of individuals with ALS show cognitive and behavioral impairments.

According to a recent review (Hobson and McDermott, 2016):

Overlap between ALS and other neurodegenerative diseases, in particular frontotemporal dementia (FTD) and parkinsonism, is increasingly recognized. ...

Approximately 10–15% of patients with ALS show signs of FTD ... typically behavioural variant of FTD. A further 50% experience mild cognitive or behavioural changes. Patients with executive dysfunction have a worse prognosis, and behavioural changes have a negative impact on carer quality of life.

This raises the issue that repetitive head trauma can result in multiple neurodegenerative diseases, not only CTE.²  In fact, this has been recognized by other researchers who studied 14 retired soccer players who were experts at heading the ball (Ling et al., 2017). Only four had pathologically confirmed CTE:

...concomitant pathologies included Alzheimer's disease (N = 6), TDP-43 (N = 6), cerebral amyloid angiopathy (N = 5), hippocampal sclerosis (N = 2), corticobasal degeneration (N = 1), dementia with Lewy bodies (N = 1), and vascular pathology (N = 1); and all would have contributed synergistically to the clinical manifestations. ...   Alzheimer's disease and TDP-43 pathologies are common concomitant findings in CTE, both of which are increasingly considered as part of the CTE pathological entity in older individuals.

So the blanket term of “CTE” can include build-up of not only tau, but other abnormal proteins typically seen in Alzheimer's disease (A_β) and the ALS-FTD spectrum (TDP-43). This lowers the utility of an in vivo marker specific to tau in diagnosing CTE in living individuals, an important enterprise because definitive diagnosis is only obtained post-mortem.

This brings us to the problematic report on Mr. McNeill's brain and the news coverage surrounding it.


CTE confirmed for 1st time in live person, according to exam of ex-NFL player

The recent study by Omalu and colleagues (2017) performed a PET scan on Mr. Neill almost 4.5 years before he died. This was before any motor signs of ALS had appeared. Clearly, 4.5 years is a very long time in the course of progressive neurodegenerative diseases, so right off the bat a comparison of his PET scan and post-mortem pathology is highly problematic.


Former Vikings linebacker Fred McNeill identified as subject of breakthrough CTE study

Another reason this study was not the “breakthrough” of news headlines is because the type of pathology plainly visible on MRI, and the type of cognitive deficits shown on neuropsychological tests, were quite typical of Alzheimer's disease and perhaps also vascular dementia. The MRI scan taken at the time of PET “showed mild, global brain atrophy with enlarged ventricles, moderate bilateral hippocampal atrophy, and diffuse white matter hyperintensities.”

Among his worst cognitive deficits at the time of testing were memory and picture naming, which is characteristic of Alzheimer's disease (AD). Likewise, the behavioral deficits reported by his wife are typically seen in AD.

Two years after the PET scan, he developed motor symptoms of ALS. His wife noted he could no longer tie his shoes or button his shirts. He developed muscle twitching in his arms and showed decreased muscle mass in his arms and shoulders. He was diagnosed with ALS 17 months prior to death, which was in addition to his presumed diagnosis of CTE.

FDA says no to marketing FDDNP for CTE

Finally, the molecular imaging probe used to identify abnormal tau protein in the living brain, [¹⁸F]-FDDNP, is not specific for tau. It also binds to beta-amyloid and a variety of other misfolded proteins. Or maybe not!

As I've written before, the brain diagnostics company TauMark™ was admonished by the FDA for making false claims. Six authors on the current paper hold a financial interest in the company. Most other research groups use more specific tau imaging tracers such as [¹⁸F]T807 (aka [¹⁸F]AV-1451 or Flortaucipir).

I certainly acknowledge that theses types of pre- and post-mortem studies are very difficult to conduct, and although the n=1 is a known weakness, you have to start somewhere. Nonetheless, the stats relating FDDNP binding to tau pathology were very thin and not all that believable. The paragraph below presents the results in their entirety. Note that p=.0202 was considered “highly correlated” while p=.1066 was not significant.

Correlation analysis was performed to investigate whether the in vivo regional [F-18]FDDNP binding level agreed with the density of tau pathology based on autopsy findings. Spearman rank-order correlation coefficient (r_s) was calculated for the regional [F-18]FDDNP DVRs (Figure 1) and the density of tau pathology, as well as for amyloid and TDP-43 substrates (Table 5). Our results showed that the tau regional findings and densities obtained from antemortem [F-18]FDDNP-PET imaging and postmortem autopsy were highly correlated (r_s = 0.592, P = .0202). However, no statistical correlation was found with the presence of amyloid deposition (r s = -0.481; P = .0695) or of TDP-43 (r_s = 0.433; P = .1066).

Also, FDDNP-PET showed that in cortical regions, the medial temporal lobes showed the highest distribution volume ratio (DVR), along with anterior and posterior cingulate cortices. Isn't this typical of the A_β distribution in AD?

I'm not denying the existence of CTE as a complex clinical entity, or saying that multiple concussions don't harm your brain. Along with others (e.g., Iverson et al., 2018), I'm merely suggesting that the clinical, cognitive, behavioral, and pathological sequelae of repeated head trauma should be carefully studied, and not presented in a sensationalistic manner.


Footnotes

¹ Illustration of the coronal plane of section.



² Note that most cases of ALS and FTD are not caused by concussions.



Read Part 1 of the series:

Brief Guide to the CTE Brains in the News. Part 1: Aaron Hernandez

References

Hobson EV, McDermott CJ. (2016). Supportive and symptomatic management of amyotrophic lateral sclerosis. Nat Rev Neurol. 12(9):526-38.

Iverson GL, Keene CD, Perry G, Castellani RJ. (2018). The Need to Separate ChronicTraumatic Encephalopathy Neuropathology from Clinical Features. J Alzheimers Dis. 61(1):17-28.

Ling H, Morris HR, Neal JW, Lees AJ, Hardy J, Holton JL, Revesz T, Williams DD. (2017). Mixed pathologies including chronic traumatic encephalopathy account fordementia in retired association football (soccer) players. Acta Neuropathol. 133(3):337-352.

Omalu B, Small GW, Bailes J, Ercoli LM, Merrill DA, Wong KP, Huang SC, Satyamurthy N, Hammers JL, Lee J, Fitzsimmons RP. (2017). Postmortem Autopsy-Confirmation of Antemortem [F-18] FDDNP-PET Scans in a Football Player With Chronic Traumatic Encephalopathy. Neurosurgery. 2017 Nov 10.


Further Reading – I've written about CTE a lot, you can read more below.

FDA says no to marketing FDDNP for CTE

Is CTE Detectable in Living NFL Players?

The Ethics of Public Diagnosis Using an Unvalidated Method

The Truth About Cognitive Impairment in Retired NFL Players

Lou Gehrig Probably Died of Lou Gehrig's Disease

Blast Wave Injury and Chronic Traumatic Encephalopathy: What's the Connection?

Little Evidence for a Direct Link between PTSD and Chronic Traumatic Encephalopathy

Brief Guide to the CTE Brains in the News. Part 1: Aaron Hernandez

Chronic traumatic encephalopathy (CTE) is the neurodegenerative disease of the moment, made famous by the violent and untimely deaths of many retired professional athletes. Repeated blows to the head sustained in contact sports such as boxing and American football can result in abnormal accumulations of tau protein (usually many years later). The autopsied brains from two of these individuals are shown below.

Left: courtesy of Dr. Ann McKee in NYT.  Right: courtesy of Dr. Bennett Omalu in CNN. These are coronal sections¹ from the autopsied brains of: (L) Aaron Hernandez, aged 27; and (R) Fred McNeill, aged 63.


Both men played professional football in the NFL. Both came upon some troubled times after leaving the game. And although the CTE pathology in their brains has been attributed directly to football — repeated concussive and sub-concussive events — other potential factors have been mostly ignored. Below I'll discuss these events and phenomena, and whether they could have contributed to the condition of the post-mortem brains.


Aaron Hernandez

Illustration by Sean McCabe for Rolling Stone

Talented ex-NFL football star, PCP addict, convicted murderer, and suicide by hanging. The Rolling Stone ran two riveting articles that detailed the life (and death) of Mr. Hernandez. Despite a difficult upbringing surrounded by violence and tragedy, he was a serious and stellar athlete at Bristol High School. The tragic death of his father from a medical accident led Aaron to hang out with a less savory crowd. He fortunately ended up at the University of Florida for college football. There he failed several drug tests, but the administration mostly looked the other way. He was on a national championship team, named an all-American, and involved in a shooting where he was not charged.

Most NFL teams took a pass because of his use of recreational drugs and reputation as a hot-head:

After seeing his pre-draft psychological report, where he received the lowest possible score, one out of 10, in the category of “social maturity” and which also noted that he enjoyed “living on the edge of acceptable behavior,” a handful of teams pulled him off their boards, and 25 others let him sink like a stone on draft day.

But he ended up signing with the New England Patriots in a $40 million deal. He smoked pot constantly and avoided hanging out with the other players. “Instead of teammates, Hernandez built a cohort of thugs, bringing stone-cold gangsters over to the house to play pool, smoke chronic and carouse.” Things spiraled downwards, in terms of thug life, use of PCP (angel dust), and ultimately the murder of a friend that ended in a life sentence without parole.

He was also tried and acquitted of a separate double homicide, but his days were numbered. Two days later he hanged himself with a bedsheet in his jail cell. He was rumored to have smoked K2 (nasty synthetic cannabis) just before his death, but this was ultimately unsubstantiated.

These complicating factors — lengthy history of drug abuse, death by asphyxiation — must have had some effect on his brain, I mused in another post.

Meanwhile, the New York Times had a splashy piece about how the pristine brain of Aaron Hernandez presented an opportunity to study a case of “pure” CTE:

What made the brain extraordinary, for the purpose of science, was not just the extent of the damage, but its singular cause. Most brains with that kind of damage have sustained a lifetime of other problems, too, from strokes to other diseases, like Alzheimer’s. Their samples are muddled, and not everything found can be connected to one particular disease.

This was a startling statement, as I said in my secondary blog:

I’ve been struggling to write a post that highlights the misleading nature of this claim. How much of that was [the writer's] own hyperbole? Or was he merely paraphrasing the famous neuropathologists who presented their results to the media, not to peer reviewers? Is it my job to find autopsied brains from PCP abusers and suicides by hanging? Searching for the latter, by the way, will turn up some very unsavory material in forensic journals and elsewhere. At any rate, I think much of this literature glosses over any complicating elements, and neglects to mention all of the cognitively intact former football players whose brains haven’t been autopsied.

In the next post, I'll discuss the case of Fred McNeill O'Neill.


Footnote

¹ Illustration of the coronal plane of section.





Read Part 2 of the series:

Brief Guide to the CTE Brains in the News. Part 2: Fred McNeill

Further Reading  
– I've written about CTE a lot, you can read more below.

FDA says no to marketing FDDNP for CTE

Is CTE Detectable in Living NFL Players?

The Ethics of Public Diagnosis Using an Unvalidated Method

The Truth About Cognitive Impairment in Retired NFL Players

Lou Gehrig Probably Died of Lou Gehrig's Disease

Blast Wave Injury and Chronic Traumatic Encephalopathy: What's the Connection?

Little Evidence for a Direct Link between PTSD and Chronic Traumatic Encephalopathy

New York Times: A neuropathologist and her associate examined slices of the brain of a 27-year-old man. Credit: Boston University.

Can a Computer Algorithm Identify Suicidal People from Brain Scans? The Answer Won't Surprise You

Just et al. 2017

Death by suicide is a preventable tragedy if the suicidal individual is identified and receives appropriate treatment. Unfortunately, some suicidal individuals do not signal their intent, and others do not receive essential assistance. Youths with severe suicidal ideation are not taken seriously in many cases, and thus are not admitted to emergency rooms. A common scenario is that resources are scarce, the ER is backed up, and a cursory clinical assessment will determine who is admitted and who will be triaged. From a practical standpoint, using fMRI to determine suicide risk is a non-starter.

Yet here we are, with media coverage blaring that an Algorithm can identify suicidal people using brain scans and Brain Patterns May Predict People At Risk Of Suicide. These media pieces herald a new study claiming that fMRI can predict suicidal ideation with 91% accuracy (Just et al. 2017). The authors applied a complex algorithm (machine learning) to analyze brain scans obtained using a highly specialized protocol to examine semantic and emotional responses to life and death concepts.

Let me unpack that a bit. The scans of 17 young adults with suicidal ideation (thoughts about suicide) were compared to those from another 17 participants without suicidal ideation. A computer algorithm (Gaussian Naive Bayes) was trained on the neural responses to death-related and suicide-related words, and correctly classified 15 out of 17 suicidal ideators (88% sensitivity) and 16 out of 17 controls (94% specificity). Are these results too good to be true? Yes, probably. And yet they're not good enough, because two at-risk individuals were not picked up.

Just et al. 2017

The computational methods used to classify the suicidal vs. control groups are suspect, according to many machine learning experts on social media. One problem is known as “overfitting” — using too many parameters taken from small populations that may not generalize to unique samples. The key metric is whether the algorithm will be able to classify individuals from independent, out-of-sample populations. And we don't know that for sure. Another problem is that the leave-one-out cross validation is problematic. I'm not an expert here, so the Twitter threads that start below (and here) are your best bet.

ML re suicide, 90% correct, 2 groups of 17. Shiny journal. Anyone see any problems ? https://t.co/mgQ8tW6s5w @tyrell_turing

— KordingLab (@KordingLab) October 31, 2017

For the rest of this post, I'll raise other issues about this study that concerned me.


Why use an expensive technology in the first place?

The rationale for this included some questionable statements.

• ...predictions by both clinicians and patients of future suicide risk have been shown to be relatively poor predictors of future suicide attempt^2,3.

One of the papers cited as a poor predictor (Nock et al., 2010) was actually touted as a breakthrough when it was published: Implicit Cognition Predicts Suicidal Behavior. [n.b. Nock is an author on the Just et al. paper that trashes his earlier work]. Anyway, Nock et al. (2010) developed the death/suicide Implicit Association Test (IAT)¹ which was able to identify ER patients at greatest risk for another suicide attempt in the future:

...the implicit association of death/suicide with self was associated with an approximately 6-fold increase in the odds of making a suicide attempt in the next 6 months, exceeding the predictive validity of known risk factors (e.g., depression, suicide-attempt history) and both patients’ and clinicians’ predictions.

But let's go ahead with an fMRI study that will be far more accurate than a short and easy-to-administer computerized test!

• Nearly 80% of patients who die by suicide deny suicidal ideation in their last contact with a mental healthcare professional⁴.

This 2003 study was based on psychiatric inpatients who died by suicide while in hospital (5-6% of all suicides) or else shortly thereafter, and may not be representative of the entire at-risk population. Nonetheless, other research shows that current risk scales are indeed of limited use and may even waste valuable clinical resources. The scales “may be missing important aspects relevant to repeat suicidal behaviour (for example social, cultural, economic or psychological processes).” But a focus on brain scans would also miss social, cultural, and economic factors.


How do you measure the neural correlates of suicidal thoughts?

This is a tough one, but the authors propose to uncover the neural signatures of specific concepts, as well as the emotions they evoke:

...the neural signature of the test concepts was treated as a decomposable biomarker of thought processes that can be used to pinpoint particular components of the alteration [in participants with suicidal ideation]. This decomposition attempts to specify a particular component of the neural signature that is altered, namely, the emotional component...

How do you choose which concepts and emotions to measure?

The “concepts” were words from three different categories (although the designation of Suicide vs. Negative seems arbitrary for some of the stimuli). The set of 30 words was presented six times, with each word shown for three seconds followed by a four second blank screen. Subjects were “asked to actively think about the concepts ... while they were displayed, thinking about their main properties (and filling in details that come to mind) and attempting consistency across presentations.”

Just et al. 2017

The “emotion signatures” were derived from a prior study (Kassam et al., 2013) that asked method actors to self-induce nine emotional states (anger, disgust, envy, fear, happiness, lust, pride, sadness, and shame). The emotional states selected for the present study were anger, pride, sadness, and shame (all chosen post hoc). Should we expect emotion signatures that are self-induced by actors to be the same as emotion signatures that are evoked by words? Should we expect a universal emotional response to Comfort or Evil or Apathy?

Six words (death, carefree, good, cruelty, praise, and trouble — in descending order) and five brain regions (left superior medial frontal, medial frontal/anterior cingulate, right middle temporal, left inferior parietal, and left inferior frontal) from a whole-brain analysis (that excluded bilateral occipital lobes for some reason) provided the most accurate discrimination between the two groups. Why these specific words and voxels? Twenty-five voxels, specifically. It doesn't matter.

The neural representation of each concept, as used by the classifier, consisted of the mean activation level of the five most stable voxels in each of the five most discriminating locations.

...and...

All of these regions, especially the left superior medial frontal area and medial frontal/anterior cingulate, have repeatedly been strongly associated with self-referential thought...

...and...

...the concept of ‘death’ evoked more shame, whereas the concept of ‘trouble’ evoked more sadness in the suicidal ideator group. ‘Trouble’ also evoked less anger in the suicidal ideator group than in the control group. The positive concept ‘carefree’ evoked less pride in the suicidal ideator group. This pattern of differences in emotional response suggests that the altered perspective in suicidal ideation may reflect a resigned acceptance of a current or future negative state of affairs, manifested by listlessness, defeat and a degree of anhedonia (less pride evoked in the concept of ‘carefree’) [why not less pride to 'praise' or 'superior'? who knows...]

Not that this involves circularity or reverse inference or HARKing or anything...


How can a method that excludes data from 55% of the target participants be useful??

This one seems like a showstopper. A total of 38 suicidal participants were scanned, but those who did not show the desired semantic effects were excluded due to “poor data quality”:

The neurosemantic analyses ... are based on 34 participants, 17 participants per group whose fMRI data quality was sufficient for accurate (normalized rank accuracy > 0.6) identification of the 30 individual concepts from their fMRI signatures. The selection of participants included in the primary analyses was based only on the technical quality of the fMRI data. The data quality was assessed in terms of the ability of a classifier to identify which of the 30 individual concepts they were thinking about with a rank accuracy of at least 0.6, based on the neural signatures evoked by the concepts. The participants who met this criterion also showed less head motion (t(77) = 2.73, P < 0.01). The criterion was not based on group discriminability.

This logic seems circular to me, despite the claim that inclusion wasn't based on group classification accuracy. Seriously, if you throw out over half of your subjects, how can your method ever be useful? Nonetheless, the 21 “poor data quality” ideators with excessive head motion and bad semantic signatures were used in an out-of-sample analysis that also revealed relatively high classification accuracy (87%) compared to the data from the same 17 “good” controls (the data from 24 “bad” controls were excluded, apparently).

We attribute the suboptimal fMRI data quality (inaccurate concept identification from its neural signature) of the excluded participants to some combination of excessive head motion and an inability to sustain attention to the task of repeatedly thinking about each stimulus concept for 3 s over a 30-min testing period.

Furthermore, another classifier was even more accurate (94%) in discriminating between suicidal ideators who had made a suicide attempt (n=9) from those who had not (n=8), although the out-of-sample accuracy for the excluded 21 was only 61%. Perhaps I'm misunderstanding something here, but I'm puzzled...

I commend the authors for studying a neglected clinical group, but wish they were more rigorous, didn't overinterpret their results, and didn't overhype the miracle of machine learning.


Crisis Text Line [741741 in the US] uses machine learning to prioritize their call load based on word usage and emojis. There is a great variety of intersectional risk factors that may lead someone to death by suicide. At present, no method can capture the full scope of diversity of who will cross the line.

If you are feeling suicidal or know someone who might be, here is a link to a directory of online and mobile suicide help services.



Footnote

¹ I won't discuss the problematic nature of the IAT here.


References

Just MA, Pan L, Cherkassky VL, McMakin DL, Cha c, Nock MK, & Brent D (2017). Machine learning of neural representations of suicide and emotion concepts identifies suicidal youth. Nature Human Behaviour. Published online: 30 October 2017

Kassam KS, Markey AR, Cherkassky VL, Loewenstein G, Just MA. (2013). Identifying Emotions on the Basis of Neural Activation. PLoS One. 8(6):e66032.

Nock MK, Park JM, Finn CT, Deliberto TL, Dour HJ, Banaji MR. (2010). Measuring the suicidal mind: implicit cognition predicts suicidal behavior. Psychol Sci. 21(4):511-7.

The Devilish Side of Psychiatry

Final Destination 3 (2006)

The devil always experienced malicious pleasure in imposing himself in neuropsychiatric nosology


Olry and Haines (2017) published a mischievous article in the Journal of the History of the Neurosciences:

Having an inquiring mind by nature, the Devil always managed to interfere in all spheres of human activity, including the sciences. ... Biologists use an enzyme called “luciferase” — Lucifer has been described as the “light-bearing” fallen angel, hence the bioluminescence — to spot certain proteins by chromogenous reactions (Lodish et al., 2005, p. 92). ...

But how did the Devil get a foot — of course cloven (!) — into the door of the neurosciences?

Demonic possession plays an important role, of course, even in modern day psychiatric nosology (see the debate over Possession Trance Disorder in DSM-5). Does it make any sense to use DSM-5 (or DSM-IV) criteria to diagnose spirit possession across cultures? Transcultural psychiatry takes a much more inclusive and sensitive approach to such phenomena, which are often precipitated by trauma.

Olry and Haines (2017) avoid this literature entirely and suggest that:

The concept of demonic possession has been mainly of theological (Omand, 1970; Balducci, 1975; Rodewyk, 1988; Amorth, 1999, 2002; Bamonte, 2006; Fortea, 2006, 2008) and/or historical concern (Villeneuve, 1975; Pigin, 1998; Kelly, 2010; Kiely & McKenna, 2007).  ...

Although conservative theologians might not question the reality of diabolical possession (see Haag, 1969; Cortès & Gatti, 1975, for the few exceptions), many psychiatrists and psychologists admit being interested in the concept though, of course, not declaring themselves in favor of a supernatural etiology...

But being diabolical sorts themselves, the authors namedrop and show off their autographed copy of The Exorcist.

Figure 1. Title page of William Peter Blatty’s The Exorcist, with signed dedication by the actress Linda Blair. Author’s (R.O.) copy.



They continue:

However, literature and the movie industry — let’s remember William Peter Blatty’s The Exorcist (Blatty, 1971) (see Fig. 1) and the sociological impact of William Friedkin’s screen adaptation two years later (Bozzuto, 1975) — not only generated impassioned movie critics ... but also brought back scientific discussions involving neurosciences and, more specifically, psychology, neurology, and psychiatry (Montgomery, 1976).

Häxan (1922) – entire film available at archive.org

Deadly exorcisms have been reported recently in the medical literature, including several cases of Fatal Hypernatraemia from Excessive Salt Ingestion During Exorcism. One 20-year-old woman received a prescription for Prozac to treat her postpartum depression, but her family also advised her to undergo an exorcism. She reportedly drank six glasses of a mixture of 1 kg table salt in a liter of water.

The Church itself involved physicians many centuries ago in the differential diagnosis between possession and mental disease, as exemplified by the 1583 Rheims National Synod:

[Before he undertakes to exorcize, the priest has to inquire diligently about the life of the possessed [. . .], of his health [. . .], because melancholics, lunatics often need much more cures of the physician than the ministry of exorcists.] (Tonquédec, 1948, p. 330)

Physicians, and in actual fact, clinical neuroscientists, then had to name a phenomenon — nosology oblige — about which most did not believe.

The Devil's Influence Over Neuropsychiatry – “some lexicological compromises”

...neuropsychiatrists sometimes allow themselves the use of theological concepts (e.g., possession, diabolical, demonological), provided that an additional term — medical or not — grants them a little more scientific credibility. This addition may be “neurosis” (demonological neurosis: Hélot, 1898; Freud, 1923), “psychosis” (diabolical possession psychosis: Lhermitte, 1944), “delirium” (diabolical possession delirium: Gayral, 1944; Delay, 1945), “syndrome” (possession syndrome: Yap, 1960), “phenomenon” (phenomenon of possession: Bron, 1975), “state” (possession state: Wittkower, 1970), or “experience” (possession experience: Pattison, 1969, p. 323).

Or sometimes the patient may feel like they are literally in hell.

Self-Portrait in Hell, by Edvard Munch (1903)

Reference

Olry R, Haines DE. (2017). The devil always experienced malicious pleasure in imposing himself in neuropsychiatric nosology. J Hist Neurosci. 26(3):329-335.


Further Reading

Possession Trance Disorder in DSM-5

Spirit Possession as a Trauma-Related Disorder in Uganda

"The spirit came for me when I went to fetch firewood" - Personal Narrative of Spirit Possession in Uganda

Possession Trance Disorder Caused by Door-to-Door Sales

Fatal Hypernatraemia from Excessive Salt Ingestion During Exorcism

Diagnostic Criteria for Demonic Possession

The Wailing (aka 곡성, 哭聲, Gokseong)

...and to make your Halloween nightmares complete...

Although it's certainly not for everybody, The Wailing is an amazing film.

BROADEN Trial of DBS for Treatment-Resistant Depression No Better than Sham

Website for the BROADEN™ study, which was terminated by
the sponsor.

A multi-site, randomized, double-blind, placebo-controlled clinical trial of deep brain stimulation (DBS) for treatment-resistant depression has failed, according to a new article in Lancet Psychiatry. The targeted brain region was bilateral subcallosal cingulate white matter, which had been called the “Depression Switch” based on acute stimulation studies at Emory. These disappointing results were not surprising, since they were covered by Neurotech Business Report in December 2013 and then in depth by my posts here and here. The new paper followed the patients for a longer period of time, up to 24 months for some in the cohort.

The main portion of the trial was six months in length. All patients received implantation surgery. Two weeks later, they were randomized to either the treatment group (n=60), who received stimulation right away, or the “sham” control group (n=30), who did not. After six months, the blinding was uncovered and both sham and treatment groups were offered open label DBS for another six months.

In the figure below, Control (red line) and Stimulation (blue line) groups both showed slight improvements over time, with no significant difference in depression severity measured by the Montgomery-Åsberg Depression Rating Scale (MADRS). This was the primary endpoint. We don't see a difference between groups at six months or any other time.

- click on image for a larger view -

Fig. 2. (Holtzheimer et al., 2017). At months 9 and 12, the control group was receiving active stimulation; therefore, for the control group, 9 months refers to 3 months of active stimulation, and 12 months refers to 6 months of active stimulation. Error bars indicate standard deviations.


Concerning the endpoint more specifically (Holtzheimer et al., 2017):

The primary efficacy endpoint for the study was defined as difference in proportion of patients achieving a response between the stimulation and control groups. Response was defined as a 40% or greater reduction in MADRS and no worsening in GAF from baseline (average of three baseline MADRS assessments) to the average scores at months 4, 5, and 6.

Table 3 (modified from Holtzheimer et al., 2017).


Here's the full scoop for the futility analysis that put an end to the trial (because of the low probability of success). I had erroneously stated in January 2014 that the trial was halted by the FDA. It wasn't. It was stopped by the sponsor, St. Jude Medical (Holtzheimer et al., 2017):

For the futility analysis, based on the first 6 months' data, the proportion of patients with response for the stimulation group was predicted to be 40%, and for the control group was predicted to be 18·5%. In the actual futility analysis, these figures were 20% for the stimulation group and 17% for the control group. It was concluded that the study had a 17% chance of success if continued. Although this did not meet the prespecified definition for futility (<10% chance of success), the sponsor chose to end study enrolment following the futility analysis.

Although “These findings are disappointing given the encouraging data from earlier open-label studies of subcallosal cingulate DBS,” all was not lost, according to the authors. They offered a number of possible explanations (which can be summarized as long duration of illness, suboptimal stimulation parameters, and lack of tractography):

• “participants in this study had an average current episode duration of about 12 years, which is much longer than the average duration of current episode in previous studies of subcallosal cingulate DBS (approximately 5 years) and might have contributed to the low overall proportion of patients achieving a response.”

• “possible that stimulation contacts and parameters were suboptimal during the first 12 months of this study, given the somewhat restrictive programming algorithm used. Greater improvement in depression occurred after the 12-month endpoint when more flexibility in stimulation contacts and parameters was allowed.”

• “Neurosurgical placement of the DBS electrodes, based on this algorithm, was highly accurate and did not differ between eventual responders and non-responders.”

[NOTE: placement was verified by at least two of three experts: HSM, CH, PEH. Nonetheless, the authors argue that placement could improve with more detailed tractography, e.g. Riva-Posse et al. 2017. This refinement of protocol has been discussed for the last 10 years; see Johansen-Berg 2007 and ...But My Subgenual Cingulate Is Sad.]

• “for maximal efficacy, the active electrode for subcallosal cingulate DBS must be placed such that it affects a crucial network of white matter tracts connecting key brain regions, including the forceps minor, cingulum bundle, and uncinate fasciculus. Therefore, it is possible that prospective targeting on the basis of individual diffusion tensor imaging tractography could optimise electrode placement in subcallosal cingulate DBS.”

In an earlier paper, a group of DBS investigators and ethics experts advised other researchers, industry mavens, and even bloggers on “Being open minded about neuromodulation trials: Finding success in our 'failures'.” (Finns et al., 2017)

“Similarly, another randomized double blind clinical trial comparing active versus sham stimulation for the treatment of severe depression targeting Brodmann Area 25 was also halted for futility prior to completion of the planned study (St Jude Medical sponsored BROADEN trial). While there are neither publications nor official industry statements, uninformed speculations as to causes of the failure are in the public domain [28, 29] to the detriment of the scientific process and progress.

In each of these instances, different combinations of variables can lead to disappointing results. For example, patient characteristics, surgical variability, stimulation algorithms, outcome metrics, and institutional variance, can all contribute to negative outcomes in complex trials that initially seem promising. Further, once a negative report is published, the work can become ‘toxic’, and there is little incentive to engage in small subset analyses that have a limited market.”

Finally,

“We believe that investigators, industry, regulators and society need to fully understand what is casually described as success and failure in order to maximize return on investment, all the more so when opportunities for additional knowledge generation remain in place. To do otherwise, would be irresponsible.”

So to call the BROADEN trial a failure is “irresponsible”? Personally, I am aware that a multi-site trial using invasive new technology to treat intractable psychiatric patients with a terrible and (ultimately) ill-defined syndrome is a massive undertaking. And very, very, very expensive. I have no problem with the investigators trying to glean what they can from individual differences to move forward with better targets/parameters/etc. I wanted to see this procedure help a majority of patients.

The bottom line here is that the primary preregistered endpoint was as follows: 12/60 (20%) improved with stimulation, 5/30 (17%) improved with no stimulation, 8/60 (12%) patients with stimulation reported an increase in depressive symptoms (this was not defined or quantified), and 1/30 (3%) patients with no stimulation reported an increase in depressive symptoms.

Let's take a look at the registered clinical trial. Oh we can't.


Clinical Trial NCT00617162

[Trial of device that is not approved or cleared by the U.S. FDA]

However, we can look at other clinical trials using the same device (Libra Deep Brain Stimulation System) with the same sponsor (St. Jude Medical) in Europe and Canada. Oh by the way, an April 2016 news release announced: Abbott to Acquire St Jude Medical (DBS was not mentioned). In January 2017 Abbott Completes the Acquisition of St. Jude Medical (no DBS here, either). I won't speculate any further. I'm too tired.


I'd like to conclude with an upbeat tweet from a prominent neuroscientist who studies pain and the placebo effect.

A needed and long awaited paper! Fascinating implications for placebo responses. https://t.co/vdCn3VXWSX

— tor wager (@torwager) October 6, 2017

References

Choi KS, Riva-Posse P, Gross RE, Mayberg HS. (2015). Mapping the "Depression Switch" During Intraoperative Testing of Subcallosal Cingulate Deep Brain Stimulation. JAMA Neurol. 72(11):1252-60.

Fins JJ, Kubu CS, Mayberg HS, Merkel R, Nuttin B, Schlaepfer TE. (2017). Being open minded about neuromodulation trials: Finding success in our "failures". Brain Stimul. 10(2):181-186. 

Holtzheimer PE, Husain MM, Lisanby SH, Taylor SF, Whitworth LA, McClintock S, Slavin KV, Berman J, McKhann GM, Patil PG, Rittberg BR. (2017). Subcallosal cingulate deep brain stimulation for treatment-resistant depression: a multisite, randomised, sham-controlled trial. The Lancet Psychiatry. 2017 Oct 4.

Riva-Posse P, Choi KS, Holtzheimer PE, Crowell AL, Garlow SJ, Rajendra JK, McIntyre CC, Gross RE, Mayberg HS. (2017). A connectomic approach for subcallosal cingulate deep brain stimulation surgery: prospective targeting in treatment-resistant depression. Mol Psychiatry. 2017 Apr 11. [Epub ahead of print].


Further Reading

BROADEN Trial of DBS for Treatment-Resistant Depression Halted by the FDA
 NOTE: the trial was actually halted by the sponsor, not the FDA

Update on the BROADEN Trial of DBS for Treatment-Resistant Depression

Deep Brain Stimulation for Bipolar Depression

Modern Tract-Tracing for Historical Psychosurgery

...But My Subgenual Cingulate Is Sad

The Sad Cingulate

Sad Cingulate on 60 Minutes and in Rats