OHDSI Home | Forums | Wiki | Github


(Ana Szarfman) #1

We propose to use the OHDSI Community Network to help us explore the natural history (NH) of the AA, AL, and ATTR forms.

  1. List item

We want to understand how best to study these diseases using clinical trial and real world evidence data.

Our goal is to increase our knowledge about the disease process to help inform design and conduct of clinical trials in these disease spaces; decide upon the best study design, select enrollment criteria, endpoints, when to measure endpoints, etc. Ultimately, we would like to bring more interest from pharmaceutical companies to these disease spaces because approved therapies are lacking. The understanding of the NH of the diseases is key.

We have many scientific questions that want the OHDSI collaborative team to help us answer by tapping their network of Real World Evidence.

• What are the prognostic factors? Ideally, we would like to query the data to see if there are any correlations between different serum biomarkers, imaging biomarkers and clinical outcomes.
• What is the correlation between demographic factors (age, sex, race, underlying medical disorders, etc.) and different presentations of the disease, treatment responses, and outcomes?
• What is the relationship between treatments and favorable or unfavorable clinical outcomes and biomarkers?
• Could a prognostic disease model be constructed using clinical characteristics +/- biomarkers +/- histology?
• Which are the typical symptoms and the most frequent and higher than expected adverse events? These diseases have very heterogeneous in symptoms (i.e., GI (nausea, vomiting, diarrhea, constipation), motor/sensory neuropathy, brain fog, anxiety/depression, cardiomyopathy with preserved ejection fraction of ≥50% (HFpEF) (usually), renal failure, skin problems, etc.
• What is the prevalence of these different clinical presentations, and how do they evolve before and after treatment?
• Are any genetic factors pathognomonic and/or predictive of progression?

(Matt Spotnitz) #2

Hi Ana,

Thank you for posting this interesting proposal. We have done some preliminary work at Columbia on this topic(@aperotte, @cukarthik ). There may be some challenges with phenotyping amylodosis using billing/claims data. First, amyloidosis is often billed as “amyloidosis” without subtype (AA, AL, ATTR) being specified. Second, a significant number of patients who have an initial diagnosis of amyloidosis may not have the disease. Amyloidosis may be listed as part of a differential diagnosis but that may not be the final diagnosis. Third, patients may be diagnosed with amyloidosis at one institution and then referred to another medical center for management. I am happy to exchange ideas with you and help you with your research objectives.


(Laura Hester) #3

Hi Ana,

I am an epidemiologist currently working with Janssen’s AL Amyloidosis Working Group. We are completing multiple projects using US EHR and claims data (converted to the OMOP CDM) to identify patients earlier in the disease course. As of today, we have (1) characterized sequences, timing, and frequencies of 26 symptoms in the clinical prodrome before diagnosis and (2) developed a clinical prediction model that could be applied to EHR data to identify potential AL amyloidosis cases earlier. We also have done some characterization of treatment pathways after diagnosis. Currently, we are working on publishing these projects. Additionally, we plan to complete the same projects in the Swedish National Registries starting this summer.

We are also starting a collaboration with Pfizer to jointly develop and validate a clinical prediction model for cardiac amyloidosis (focused on AL and ATTR).

As Matt mentioned, the phenotype for amyloidosis is difficult, especially AL and ATTR. Two ICD-10 codes were introduced for these amyloidoses in late 2017, and as of our most recent data update, we only have about 130-150 people with 2+ ICD-10 E85.81 codes for AL amyloidosis and 365 days of prior continuous observation. We also have a phenotype for AL amyloidosis based on prior publications that we are working on validating. This phenotype relies on identifying patients who receive a multiple myeloma treatment after receiving an algorithm for “other amyloidosis”, and therefore, only represents treated patients (but has a population of closer to 1400 patients since 2001 in Optum claims data). The treatments used follow the recommendations in the NCCN guidelines for AL amyloidosis.

We also have interest in data that has more information about biomarkers and genomics. We would appreciate the opportunity to collaborate with others who might have insight on these topics.

I’m happy to work with you and others on amyloidosis, and if there is interest, find a way to share the results from our current studies.



(Jeremy Warner) #4

Hi Ana,

Sam Rubinstein is a senior hematology/oncology fellow here at Vanderbilt with a specific interest in amyloidosis. He is not on the OHDSI forum but is very interested in learning more and potentially collaborating with you. He can be reached at samuel.m.rubinstein@vumc.org


(Kristin Kostka, MPH) #5


Happy to contribute our IQVIA data to your analyses. I might suggest @mattspotnitz coordinates a quick call between you, @lhester Sam and whoever else raises their hand on this thread. Sounds like there could be an OHDSI Amyloidosis WG in the making.

Btw, who is this “we” you mention? :wink:


(Leanne Goldstein) #6

I would like to contribute to this. Please include me and I can also help with coordination if needed. So great to see we are tackling this area.

(Ana Szarfman) #7

Wonderful!!! Many thanks for all your responses. I’ll be back at the office on Mon and discuss them with colleagues that are the experts in this orphan disease area.

(Ana Szarfman) #8

Thanks Kristin.

You asked: “who is this ‘we’ you mention?”

We are Preston Dunnmon(1), Melanie Blank(2), Frank Weichold(3), Gregory Pappas(4), and Ana Szarfman(1)