Thread by @jesusrbano

My comments on MERINO (thread) (1/12) on @JAMA_current 1st, CONGRATS @padstamundo @davidantibiotic and coauth - a big achievement. Disclosures: I've published papers in favor of BLBLI as alternative to carbapenems; I am @davidantibiotic friend. I was member of the DSMB of MERINO

2/12. Primary outcome was 30-day mortality. Problem: ALL deaths were unrelated to the infection but due to underlying condition, mostly metastatic/terminal cancer (Table e6). Differences in secondary endopoints were not significant (but on the side of favouring meropenem)

3/12 Rates of metastatic/terminal cancer in both arms (main cause of death) not provided. Might be disbalanced, as were UTI (12.2%!! difference, p=0.01) and qSOFA (p=0.03). Important: cancer is not included in APACHE-II or Pitt, which may not be appropriate for this population

4/12 Also, at least half mortality in piptaz ocurred afte day 15 (eFigure 4). By the way, this figure is visually misleading (Y-axis should go from 1 to 1.0, not 0.25).

5/12 No association or trend between piptaz MIC and mortality. If MIC were reliable (they might ot me, see below), this is also against the primary endpoint (30-day mortality in this population) as a measure of the effect of the antibiotic

6/12 Only 80% isolates available for a centralised analysis. A problem as (a) in those checked, resistance to assigned drug was 4.5% in piptaz vs 0.5% in mer (p=0.04); (b) there is a warning by EUCAST about the use of stripts for piptaz eucast.org/ast_of_bacteri… ...

7/12 ... (c) 60% of isolates were OXA-1 and 10% AmpC - higher resistance rates to piptazo would be expected, so appropriate AST (microdilution or disk) is needed. It can be aurgued that this is real life - but then problem would be the testing of susceptibility, not the drug.

8/12 Site effect: mortality was much higher in piptaz arm in 3 sites (the 3 from Turkey and South Africa, eTable 4). Just excluding these, mortality in piptaz arm decrease from 12.3% to 9.4%.

9/12 The multivariate analysis performed (Table 5) may not have controlled confouding (yes, despite being a RCT, due to random disbalance). Covariables to include: site effect (recruited in the above 3 sites or any other); UTI; terminal/metast cancer).

10/12 The piptazo arm of propensity-matched pairs in INCREMENT (availale in suppl table S5 at aac.asm.org/content/aac/su… ) showed pretty similar features as in MERINO, but mortality was only 7.3%. I hypothesize the proportion of advanced cancer is higher in piptazo arm of MERINO

11/12 Conclusions: (a) I don't think causality of piptazo on mortality is demonstrated in MERINO as it was unrelated to infection and there might be disbalance in key variables (b) MICs should be reassured.
These trials are a challange to perform and analyse/interpret; and...

12/12 For piptazo, I am worried about the trend favouring mer in secondary endpoints; replication of results would be needed. Meanwhile I will still use piptazo in some patients. Acknowledgements: Belén Gutiérrez, Gunnar Kalhmeter, @DrToddLee for insighful comments