Richard Lehman’s journal review—5 December 2016

richard_lehmanNEJM  1 Dec 2016  Vol 375
Can genes prove how drugs work?
Medicine is the application of neat science to a messy world. We love it when it works simply: for example, when a single gene controls a single biochemical process, which we can then block with a single chemical. Statins are often cited as an example: they block 3-hydroxy-3-methylglutaryl–coenzyme A reductase, an enzyme that is governed by the HMGCR gene. By doing so, they reduce circulating levels of low density lipoprotein cholesterol (LDL-C) and, bingo, down goes your cardiovascular risk. But for less obvious reasons, they also raise blood glucose slightly and cause muscle symptoms in many people. If your glucose is highish to start with, they may cause you to “get diabetes,” i.e. cross an artificial threshold and bundle you together with millions of different people with this scary label. But can all this really be put down to the HMGCR gene? Here’s a population study, which concludes that it can, provided you throw in a second gene locus governing LDL-C levels.
This is the PCSK9 gene, which controls proprotein convertase subtilisin–kexin type 9 expression. If you look at natural variants of these two gene loci in a large population, you’ll find exactly what the theory predicts: the level of gene expression correlates with the level of cardiovascular risk, which correlates with LDL-C level, and there is a small elevation in blood glucose in those on the “diabetic” borderline. So does this mean that PCSK9 inhibitors will inevitably prove as or more effective than statins for long term cardiovascular protection in symptomless adults? The science is neat but the world is messy. We may know for sure in 10-15 years. Oh, and I mentioned muscle effects. Maybe it’s high time to look prospectively for those, in both drug classes. And this study needs replication in other populations. Medical science is actually never simple.

JAMA  29 Nov 2016  Vol 316
Two is as good as three for HPV
Although vaccination has done more good to the world than any other medical intervention, it’s still important to treat each new mass vaccination programme as a human experiment on a large scale. Once again, the science is simple but the world is complex. In the case of vaccination against human papillomaviruses transmitted sexually and associated with anal, cervical, and oropharyngeal cancers, we are awaiting the long term reductions in these conditions, which ought to follow elimination of the viruses; and the results are very promising. No adverse effects have emerged when teenage populations have been vaccinated, whereas viral transmission has fallen even more dramatically than expected. The question of how many doses are needed is now also practically settled: this trial across 52 sites in 15 countries confirms that two shots of 9-valent vaccine provide lasting immunogenicity. This is already standard practice in a number of countries and the accompanying editorial gives a good summary of the current position.

Getting AI to eyeball the boring stuff
Pattern recognition is a big part of medicine. Some doctors, like radiologists and histopathologists, spend their whole working lives using their clever eyes and brains to convert visual images into diagnostic reports, and that is just what we humble generalists need in order to get anything done. But increasingly, where there’s a pattern, there’s a machine that can recognise it better than a human, using deep learning algorithms. The day may be fast approaching when the diagnostic “gold standard” for many routine kinds of image reporting won’t be a panel of experts but a superior machine. Here’s a study of one that reads retinal photographs for signs of diabetic retinopathy. The data set was based on over 110 000 retinal images taken in several countries using different cameras, so there is a lot of complex data in this report. But overall, it looks as if the automated machine with its “deep neural network” learning system achieved sensitivity and specificity levels at least as good as any single human reader. Compared with a panel of at least seven ophthalmologists for each image across the whole dataset, the algorithm could be ratcheted up to achieve 98% levels for either, according to how you set it. In the future it seems very likely that ophthalmologists will be trained and assessed by machines, and not the other way round.

JAMA Intern Med  Nov/Dev 2016
Financial cancer eats oncology
A couple of weeks ago I vented my rage on oncology drugs in response to an excellent BMJ article by Peter Wise. I said then that the situation could not possibly be worse, but not much more than a week later it did get worse with the passing of the 21st Century Cures bill in the US. Here is a short study of 18 new oncology drugs already given expedited approval by the US Food and Drug Administration (FDA). Of these, 13 were priced above $100K annually, and the most expensive, cabozantinib, did not improve overall survival and reduced quality of life compared to placebo. Thank goodness that we have Vinay Prasad and Bishal Gyawali blazing away at this, because it looks as if we can forget about any kind of regulation by the FDA. If you want to become a billionaire, find some plausible site on a cancer cell, make an antibody, think of a funny name, and stick “uzumab” on the end of it. Run your own trial—open-label, compared with anything you like—choose a time point and an outcome where it appears to have some effect, and then sell it at the market price.

Goal driven care for dementia
What a relief to go from that to something that is simple, free, and helps people to die more comfortably. Dying slowly from dementia in a nursing home is a common alternative to dying from cancer or cardiovascular disease. Many of us, dear readers, will end our lives this way. Hopefully, we’ll have lost awareness, but our loved ones will come to us day by day, hoping for a flicker of recognition, and wondering how this will end, and how soon. At least they should know that there is a care plan that takes into account their goals and wishes, and I think this is now pretty universally the case in the UK. In case you need trial evidence that this is not just the right thing to do, but also achieves better outcomes in the US, here it is. The goals of care decision aid they tested resulted in better communication, more involvement from palliative care, and fewer hospital admissions.

Lancet  3 Dec 2016  Vol 388
A FALCON for fulvestrant
Fulvestrant is an old drug, which works by degrading the oestrogen receptors in receptor-positive breast cancer. Its original patent ran out in 2004, but AstraZeneca managed to get a new one that expires in 2021. In the UK, treatment with fulvestrant (Faslodex) by monthly injection costs £522, compared with anastrozole daily by mouth which costs £6 per month. Here is an AstraZeneca funded trial called FALCON, which concludes that “Fulvestrant has superior efficacy and is a preferred treatment option for patients with hormone receptor-positive locally advanced or metastatic breast cancer who have not received previous endocrine therapy compared with a third-generation aromatase inhibitor, a standard of care for first-line treatment of these patients.” All-cause death is mentioned as a primary outcome in the abstract, but only progression-free survival is mentioned in the summary of findings, with a modest improvement of less than three months.

Omecamtiv mecarbil
The mellifluous sound of this new heart failure drug reminded me irresistibly of “O my darling Clementine,” and I felt a song coming on. But when I looked up the lyrics about Clementine, they turned out to be mighty sad, whereas omecamtiv mecarbil is meant to gladden the heart. It is a cardiac myosin activator. In a phase 2 trial in 299 patients with systolic heart failure recruited from 87 sites in 13 countries (!), omecamtiv mecarbil given for 20 weeks improved systolic function, reduced ventricular diameter, and lowered circulating levels of B natriuretic peptide. A good start, then, though we have seen many inotropic drugs come and go previously.

O me camtiv, o me camtiv, omecamtiv mecarbil!

There must be a song in this somewhere. I shall save it for the phase 3 trial. Or perhaps by then she will be lost and gone forever.

The BMJ  3 Dec 2016  Vol 355
Do alpha blockers unblock stones?
Two years ago, a Cochrane review of α-adrenergic blockers for people with ureteric stones looked at 38 small and not very good randomised trials and concluded that they were worth prescribing and might double the chances of the stone passing spontaneously. Then, a year later, a large randomised trial with good methodology found that they had no effect. Now a new systematic review and meta-analysis, including this trial and another 23 in addition to the ones in the Cochrane review, concludes that α-blockers do improve the chance of spontaneous stone expulsion, particularly if the stone is large. So while in general I would say that a well conducted RCT with a clinically important endpoint trumps a meta-analysis of many weaker trials with purely imaging based endpoints, these authors argue the opposite. Welcome to the whacky world of evidence based medicine. My evidence is better than yours. No it isn’t. Look at my forest plot. No, look at mine, it’s got more trials on it. Somehow I don’t think I’m volunteering to make a shared decision aid on this topic.

Testosterone for clots
The world’s most popular story for one thousand years was about testosterone driven men behaving like clots. This is just my plug for you to put a good version of Gilgamesh on your Christmas list. Now let’s move on to the evidence about whether giving men testosterone produces clots in their leg veins. The article begins by telling us that sales of testosterone for men have increased tenfold in the US over the past decade, which could account for a lot. But then they’ve increased 40-fold in Canada, and they elected the right leader. Anyway, back to the clots. In the absence of large randomised trials, the investigators did a population based case-control study of all men aged 20 to 89 years on the UK Clinical Research Database between 1 January 2001 and 31 May 2013. They looked at the association between testosterone use and the risk of venous thromboembolism in men, focusing particularly on the timing of exposure to testosterone and the presence of risk factors for venous thromboembolism. And they found that starting testosterone treatment was associated with an increased risk of venous thromboembolism, which peaked within six months and declined thereafter. This makes sense, even though it isn’t the highest form of evidence.

Shock horror: docs cause cancer
South Korea has the highest incidence of thyroid cancer in the world, and doctors are to blame. One lot of doctors go looking for asymptomatic thyroid nodules using ultrasound, another lot remove them, and a third lot slice and stain them and pronounce that they are cancerous. I guess all the individuals concerned then become uninsurable, though very few actually have anything resembling true cancer, since the histological diagnosis of thyroid malignancy is a mess. In fact, the mortality rate from thyroid cancer has remained level in Korea, and when you compare the survival of these “cancer patients” to the population average, it is more than 100%. That is because people seeking screening are wealthier and healthier. Nobody should mind this. It has increased economic activity in South Korea, and great wealth will continue to accrue from selling people unnecessary tests and treatments. If you are that intending billionaire I mentioned earlier, I would add genomic testing to your thyroid ultrasound and cancer drug portfolio. Behold, darkness shall cover the earth, and gross darkness the people.

Plant of the Week: Mahonia japonica
Like quite a few plants called “japonica,” this mahonia did not originate in Japan, but came West from there in the early 19th century. It first sprang up on a much warmer island off the coast of China: Taiwan. This makes its absolute toughness in the British climate something of a miracle.

We’ve had one on a dry shady bank for ages. It stands there imperturbably, its leaves prettily covered with frost and its sprays of yellow flower poised to open further when the ice melts. On warm days they will spread out their scent, like lily of the valley but perhaps even sweeter.

Although our garden is much too small for duplicate large shrubs, we bought another one this year. For years we’ve been trying to hide our oil tank from the outlook of our kitchen window, but the choice shrubs we picked all died. Digging down, I found that there is a solid concrete platform about half a metre below ground, so the roots rot and die. So far, the mahonia has shown no fear. It has grown prodigiously, and now half hides the tank. It is a lovely thing to look at too: not as dainty as some, but still a splendidly oriental presence in a dank English winter garden.