Alex Nowbar’s journal review—5 August 2019

Alex Nowbar reviews the latest research from the top medical journals

JAMA Internal Medicine
Haematuria by the book
Georgieva et al analysed the outcomes of following different guidelines for investigation of gross haematuria using a hypothetical cohort of 100 000 people. The hypothetical cohort was based on published literature about real people. They calculated rates of missed urinary tract cancers via five different sets of guidelines, along with the harms and costs of investigations like ultrasound, CT urography, and flexible cystoscopy. Guidelines that more intensively investigated found more cancers but at the expense of more CT radiation-induced cancers and more cost. Is it acceptable to miss as many cancers as the least invasive guideline (for example, omitting imaging in low risk patients, and ultrasound in preference to CT)? Based on this study’s estimated costs and harms, it probably isn’t worth pursuing a more aggressive pathway. Acceptability also depends on the consequences of missed detection compared with detected cases because missed cancer doesn’t always equate to worse health outcomes. This study doesn’t report outcomes beyond the detection of a case of cancer so the conclusions we can draw are limited. It certainly opened my eyes to the challenges of a screening programme involving CT scans though.

Acupuncture punked
In a randomised sham controlled trial of patients with “chronic stable angina,” acupuncture of the heart meridian (two points on the wrist) reduced the frequency of angina episodes compared with acupuncture of the lung meridian (two points on the arm) and compared to acupuncture of non-acupoint areas (a different two points on the arm). There is a strong temptation to dismiss this result simply because it doesn’t fit with existing beliefs. But even if you think meridians are akin to magic, and the idea that inserting needles deep into the body’s “energy highways” does anything other than induce a placebo effect (and hurt) offends you, I would concede that acupuncture could have an effect on angina.

Acupuncture researchers should also make some concessions though: first, there isn’t really such a thing as truly sham acupuncture. The patients are conscious of what they are having, and indeed the consciousness of the procedure is likely to be part of its impact. A placebo effect cannot be ruled out. A test of blinding (which is surely obligatory for sham controlled trials) might have helped.

Second, this trial does not report any of the patients’ imaging or angiographic findings, which one might use to characterise a patient with angina. This information is critical because it is the concept that needles in the arm somehow overcome ischaemia and/or coronary artery disease that causes consternation. Whether you believe the finding of efficacy of acupuncture or not, we have no idea what patient population it applies to. Third, it is unclear if acupuncture would need to be continued indefinitely to maintain the angina relief. This would be unfeasible and expensive.

Detecting atrial fibrillation from an ECG in sinus rhythm
Attia et al trained a convolutional neural network to detect atrial fibrillation on standard 12 lead ECGs showing sinus rhythm. Their methodology appears sound and well justified. Their network accurately predicted recent atrial fibrillation, which suggests that AF remodelling of the heart produces subtle features on a 12 lead ECG. The authors draw an interesting comparison of their new test for AF when currently in sinus rhythm with HbA1c for diabetes even when the patient has a normal fasting glucose. Like pre-diabetes, will we start hearing about pre-AF soon? For the patients’ sake I hope not, or at least not until we have interventions that might change outcomes.

The main application of these results as they stand would be to stratify patients after a stroke who are in sinus rhythm into those who should receive anticoagulation but this would need to be backed up by evidence from a randomised controlled trial first. Certainly, it could be more efficient than Holter monitoring and loop recorders after a stroke. It should not be used to screen asymptomatic patients more broadly because we don’t have data that looking for and finding atrial fibrillation is a good thing to be doing for the population—despite a growing lobby for this.

Chronic leukocytic leukaemia therapy
Shanafelt et al randomised (in an unblinded fashion) people with previously untreated chronic lymphocytic leukemia to the new ibrutinib–rituximab regime or to standard chemoimmunotherapy (fludarabine–cyclophosphamide–rituximab). The primary endpoint was progression-free survival. This interim analysis shows that the new regime was very effective with a hazard ratio of 0.35 at three years. The hazard ratio for overall survival was a remarkable 0.17 in favour of ibrutinib–rituximab with the caveat that there were very small numbers of deaths (four compared to 10 in the standard chemoimmunotherapy group). Adverse event rates were similar. This isn’t enough evidence for ibrutinib–rituximab to replace the standard chemoimmunotherapy though because the longer term follow-up of this study is still pending. Ibrutinib is very expensive and with some significant adverse effects, so the ultimate bar for this to enter clinical practice will be high.

CDK4/6 inhibitor for breast cancer
MONALEESA-7 is an ongoing double blind randomised controlled trial of ribociclib for advanced hormone-receptor–positive, HER2 negative breast cancer in patients who are having endocrine therapy. This interim analysis shows positive results for overall survival. I do not know why people publish these interim results—the numbers are so small at long term follow-up that it’s not meaningful to measure the difference. It’s unscientific and ostentatious with next to no implications for clinical practice or researchers. This seems to be a common oncology phenomenon. I look forward to the final results for both of these studies.

Annals of Internal Medicine
Cardiac device complication rates
Ranasinghe et al analysed complications from pacemakers and implantable cardioverter defibrillators (ICDs) in 174 institutions across New Zealand and Australia, covering more than 6000 patients up to 90 days after discharge, from 2010 to 2015. The median risk-standardized complication rate was 8%. This is quite high given that implanting these devices is considered to be a fairly routine procedure. It is useful to be reminded that undertaking these procedures should not be taken lightly. The risk-standardized complication rate ranged from 5% to 14% between institutions. The variation was still present even when only considering elective procedures, which suggests that there are some institution factors accounting for the differences. However, regional patient factors cannot be excluded despite the risk standardisation used here. 

While ICD complication rates were higher than pacemaker complication rates, many more pacemakers are performed and pacemakers accounted for three quarters of total device complications. There are measures that can reduce complication risks so perhaps some local policies require adjustment because there is clearly room for improvement at some institutions.

SGLT-2 inhibitors and UTIS
Empagliflozin. Dapagliflozin. Canagliflozin. I have learnt two important things about the gliflozins this week. Firstly, the pronunciation requires emphasis on the “glif.” Well done if you were already saying it correctly. If you weren’t, also well done, you’re on the same page as the rest of us. Second, Microsoft Word spellchecks the gliflozins correctly. This is troubling for me; medical spelling was my niche. How did they get that in the dictionary?

OK, I learned a third thing too. People are still anxious about the gliflozins. I don’t like drugs containing a z any more than the next person, but according to this large observational study they aren’t associated with severe urinary tract infections (UTIs). This study used propensity matching to test the drug’s association with UTIs bad enough to be hospitalised or septic. The comparator groups were GLP-1 agonists and DPP4-inhibitors. The authors used commercial claims databases, which have the disadvantage of representing a particular cohort of the population but the advantages of completeness and a large sample size.

Filgotinib for rheumatoid arthritis
To assess the benefit of tyrosine kinase inhibitor filgotinib for treatment of rheumatoid arthritis that has not responded to disease modifying anti-rheumatic drugs, this double blinded randomised controlled trial looked for a 20% improvement in American College of Rheumatology criteria at 12 weeks. This was achieved in 31% of people who received placebo, in 66% of people who received the 200mg dose of filgotinib, and 58% of people who received the 100mg dose. This difference is both statistically and clinically significant. All that remains is for longer term safety and efficacy to be assessed.

Sugar control after stroke
Wards are full of people with diabetes and many factors can contribute to them becoming hyperglycaemic despite their usual treatment. When the numbers are high, we bring them down, as we do with most measured things in hospital, even though there isn’t much evidence for it. In this large randomised controlled trial, people admitted with acute ischaemic stroke and hyperglycaemia were randomised to intensive or standard control for 72 hours. Intensive control was intravenous insulin infusion aiming for glucose between 4.4-7.2 mmol/L whereas standard was a subcutaneous insulin sliding scale aiming for 4.4-9.9 mmol/L. There was no difference in the proportion of people with a favourable functional outcome between the two groups. 

This trial is a little bizarre to me because the treatment targets of the two groups are so close. They both seem to have had extremely intensive glucose lowering—certainly compared to what I have seen done in any acute medical setting in the NHS.

Alex Nowbar is a clinical research fellow at Imperial College London.