Thread by @NephJC

Welcome to the @NSMCInternship #tweetorial for this week’s study on a new risk prediction tool for IgA nephropathy, appearing in @JAMAInternalMed. Check out these unhappy glomeruli with mesangial IgA deposits in green from pathologyoutlines.com (1/16)

Recent unsuccessful #RCT: STOP-IgA (immunosuppression+supportive care isn’t superior to supportive care alone) & TESTING (corticosteroids reduce risk of #ESRD but cause serious infections) were covered by @NephJC: nephjc.com/iga-nephropath…
and nephjc.com/news/2017/8/28… … (2/16)

Question for the #NephTwitter: What percentage of patients with IgA nephropathy develop #ESKD by 10 years? (3/16)

10-year risk of #ESRD ranges from 5% to 60%?!? The ability to predict decline based on clinical, histological and laboratory parameters would be invaluable. Here’s a nice review in @DiseasePrimers about some of these aspects and more: nature.com/articles/nrdp2… (4/16)

Cue Barbour and colleagues’ study published in April in @JAMAInternalMed. The authors combined clinical, histological and demographic factors to create a tool to predict the risk of #ESKD or loss of 50% of GFR at the time of biopsy jamanetwork.com/journals/jamai… … (5/16)

The tool was built using data from patient cohorts (18 years or older) from Europe, the Americas and Asia who had idiopathic IgA nephropathy proven by MEST-scored biopsy and follow-up of at least 5 years. Here’s a breakdown of the cohorts (6/16)

Different predictive models were used to construct 2 models: presented as web-based and app-based calculators. The web tool is available here: qxmd.com/calculate/calc… …, and was made in two parts: derivation & validation. Here’s a breakdown of the two cohorts below (7/16)

Both derivation and validation cohorts had similar number of cases reach the primary outcomes of #ESKD or loss of 50% GFR from time of biopsy. The five-year risks were 14.7% and 13.3% for the derivation and validation cohorts respectively (8/16)

Derivation first. Integration of clinical/histological/demographic info made the model better at predicting primary outcome risk. The different parameters were used to assess the performance of the tool are summarised in this @NephJC post: nephjc.com/news/igarisksc… … (9/16)

Next comes the validation of the models. When 1146 patients were used to validate the models, good agreement between predicted and observed risk of #ESKD or loss of 50% GFR was observed (10/16)

Finally, the tool was used to stratify patients into subgroups with increasing categories of risk. The lowest risk group only had a 1.5% mean predicted of #ESKD or loss of 50% GFR by 5 years, whereas the highest risk group had a startling 46.5% mean risk (11/16)

Here’s a challenge. A 48yo Caucasian male is referred to you with an eGFR of 75 and 1.5g/day proteinuria. His BP is 120/76. Biopsy shows some mesangial and endothelial hypercellularity. What is his risk of >50% decline in eGFR after 5 years? (12/16)

According to the IgA tool, the answer is <5%. Next scenario: this time a 12yo Chinese female with an eGFR of 43, 2.7g/day proteinuria, BP 140/97, with evidence of segmental glomerulosclerosis on biopsy. What’s her 10-year risk? (13/16)

The IgA tool won’t give a risk here, as the study didn’t include patients younger than 18 and didn't include follow-up longer than 5 years! Some other limitations are described nicely in this balanced commentary by Prof John Feehally: nephjc.com/news/2019/5/26… … (14/16)

And finally, looking forward, are there any other approaches that could predict long-term decline even better in IgA nephropathy? Could sophisticated #MachineLearning and #ArtificialIntelligence approaches predict factors we don’t even know about yet? (15/16)

And that’s a wrap! Thanks for checking out this @NSMCInternship #tweetorial by @daniyal_jafree and stay tuned for the next. Keep up to date with the latest @NephJC updates by following the account or keeping an eye on our hashtag #NephJC (16/16)