The thymosin-α1 evidence in hepatitis B is too weak to act on

Design

Cochrane systematic review and random-effects meta-analysis, RoB 2 risk of bias assessment, GRADE certainty grading

Population

10 randomised trials, 1,349 participants with chronic hepatitis B across six countries, published 1991–2018

Primary outcome

all-cause mortality, serious adverse events and health-related quality of life

Effect

all-cause mortality RR 0.53 (95% CI 0.29–0.96, very low certainty); serious adverse events RR 0.72 (0.53–0.99, low certainty); histological improvement RR 0.51 (0.13–2.06, I² = 74%)

Thymosin-α1 is used as an immunomodulator in chronic hepatitis B in several Asian countries, on the strength of trials that have never agreed with each other. This Cochrane review assembled all of them: 10 randomised trials from Bangladesh, China, Italy, Korea, Singapore and Taiwan, published between 1991 and 2018, with 1,349 participants, follow-up from six months to five years.

The pooled numbers look encouraging. All-cause mortality risk ratio 0.53 (95% CI 0.29–0.96) across three trials and 907 participants, with no heterogeneity. Serious adverse events 0.72 (0.53–0.99). Non-serious adverse events 0.47 (0.27–0.83). Quality of life and histological improvement showed little or no effect, and hepatitis B-related morbidity was uncertain (0.86, 0.54–1.40).

Then the grading, which is the review. Certainty was very low for every outcome except serious adverse events, which reached low. The reasons were high or unclear risk of bias in the trials themselves, imprecision from small numbers and intervals brushing 1.0, and inconsistency. Sixteen further studies are awaiting assessment because they were reported too incompletely to use, and the authors' enquiries went unanswered. The honest conclusion is not that thymosin-α1 does not work — it is that after 35 years of trials nobody can say.

• Do not add thymosin-α1 to a nucleoside or nucleotide analogue on the strength of these numbers.
• Entecavir and tenofovir remain the treatments with mortality and morbidity evidence behind them.
• The trials are old — published 1991 to 2018 — and most predate current standard therapy, so the comparator often is not what you would use today.
• Sixteen unusable studies with unresponsive authors is itself a finding about the literature in this area.
• Thymosin-α1 is available and prescribed in parts of Asia including India; this is the evidence base behind that practice.

Why it matters

A drug in routine use in several countries turns out to rest on evidence Cochrane cannot grade above very low.

Don't overread it

Very low certainty means the review cannot say the drug does not work either — this is absence of evidence, not evidence of absence.

The statistics, in plain English

A risk ratio of 0.53 for mortality with an upper bound of 0.96 looks convincing until you read the GRADE assessment. Very low certainty means the true effect could be substantially different in either direction — the number is not a best estimate so much as what these particular flawed trials happened to produce. An I-squared of 0% means the trials agreed with each other, which is often mistaken for reliability; trials sharing the same design flaw agree perfectly. Three trials and 907 participants is a small evidence base for a mortality claim in a disease this common.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.