Yes — the test you’re probably remembering was PGxPredict:RITUXIMAB, marketed by PGxHealth (a division of Clinical Data, later tied to Clinical Data/PGxHealth branding). It was launched in 2007 as a pharmacogenetic test meant to predict response to rituximab in follicular non-Hodgkin lymphoma, based on the FCGR3A 158V/F polymorphism.
This also fits your memory of Genentech getting unhappy. In its 2008 citizen petition to FDA,
https://www.aab.org/images/aab/pdf/Genentch%20FDA%20Petition.pdf
https://www.ncbi.nlm.nih.gov/books/NBK220030/ Natl Acad Sci
2014 DNA Cell Biol LIU 158VF and Ritxuimab and MetaAnalysis
Genentech argued that predictive LDTs used to make therapeutic decisions should face FDA review, and the petition specifically discussed PGxPredict:RITUXIMAB as an example. Genentech objected to claims that the test could identify which lymphoma patients were more likely to respond to Rituxan, arguing the supporting data were not rigorous enough. The National Academies workshop summary quotes Genentech’s representative criticizing the assay and notes that the maker claimed it could let physicians “confidently predict” rituximab response.
So while I can’t prove from the sources in front of me that FDA took a direct enforcement action against that specific assay, the test that best matches your description is PGxPredict:RITUXIMAB. The broader dispute was important enough that it became part of the public debate over LDT oversight, predictive claims, and drug-diagnostic co-development.
PGxHealth basically ceased to exist as an independent diagnostics business when its parent, Clinical Data, changed strategy. In late 2010, Clinical Data sold PGxHealth’s genetic testing and pharmacogenomic biomarker business — including the FAMILION line and reimbursement contracts — to Transgenomic for about $15.5 million. Clinical Data said explicitly that this sale completed its transformation into a pharmaceutical company.
At about the same time, PGxHealth, LLC itself was effectively repurposed on the pharma side. SEC filings show that Trovis Pharmaceuticals LLC was formerly known as PGxHealth, LLC, which tells you the brand/business was being folded into Clinical Data’s drug-development effort rather than continuing as the old pharmacogenomics testing business.
Then, in 2011, Forest Laboratories acquired Clinical Data for about $1.2 billion, mainly to obtain Viibryd (vilazodone) and the associated pharma pipeline. By that point, the original PGxHealth diagnostics operation had already been sold off, so what remained was absorbed into the broader pharma transaction.
So the short historical arc is: Genaissance assets → PGxHealth under Clinical Data → diagnostics business sold to Transgenomic in 2010 → remaining pharma-focused parent sold to Forest in 2011. In other words, PGxHealth did not disappear because one famous test failed; it disappeared because its parent company pivoted away from diagnostics and toward therapeutics, then got acquired.
If you want, I can also trace what became of the specific PGxHealth tests like FAMILION or PGxPredict:RITUXIMAB after the Transgenomic deal.
Yes. The short version is:
FAMILION lived on; PGxPredict:RITUXIMAB seems to have faded out.
After Clinical Data sold PGxHealth’s diagnostics business to Transgenomic in late 2010, the acquired assets explicitly included both the FAMILION inherited-cardiology tests and the biomarker assets behind the PGxPredict drug-response tests. Transgenomic then publicly promoted the FAMILION line in 2011, which suggests that the cardiogenetics business was the more durable, commercially central piece of what it bought.
By contrast, I do not see evidence that PGxPredict:RITUXIMAB became an enduring commercial product after the sale. I can confirm it was launched in 2007 and remained well known enough to be cited in later policy and reimbursement literature as an example of a PGx test, but I am not finding strong evidence of continued market presence or later relaunch under Transgenomic. That usually means it was either de-emphasized, quietly discontinued, or left as legacy IP rather than a growth product. I would treat that as the most likely interpretation, though I can’t prove the exact discontinuation date from the sources I found.
The corporate chain after that was fairly straightforward. Transgenomic later merged with Precipio in 2017, so Transgenomic itself did not remain a long-term standalone home for these assets.
So, practically speaking:
FAMILION / inherited cardiology tests
These were the part that clearly had continuing commercial life after the PGxHealth sale. They were integrated into Transgenomic’s specialty diagnostics business and actively expanded in 2011.
PGxPredict:RITUXIMAB
This appears to have become more of a historical footnote in the LDT / companion-diagnostic debate than a successful long-lived product line. It remained cited in the literature, but I do not see evidence it matured into a durable commercial franchise.
So the answer is not that both lines vanished overnight. It is more that the inherited disease testing franchise survived, while the rituximab-response assay seems to have withered commercially.
I found clear evidence for the 2010 sale, the continuation of FAMILION, and the later Transgenomic-Precipio merger, but not a definitive public notice saying “PGxPredict:RITUXIMAB was discontinued on X date.”
Yes — that 2014 Fen Liu paper is best read as a substantially negative or deflating follow-up to the earlier enthusiasm around FCGR3A 158V/F as a predictor of rituximab benefit in DLBCL.
The earlier biological idea was attractive. FCGR3A encodes FcγRIIIa, an Fc receptor involved in antibody-dependent cellular cytotoxicity (ADCC), and rituximab is an anti-CD20 antibody. The 158V variant was thought to bind IgG more avidly than 158F, so it was plausible that patients with V-containing genotypes might get more immune-mediated benefit from rituximab. That hypothesis was supported by an influential 2006 Korean study of 113 R-CHOP patients, which reported markedly higher complete response rates in V/V patients than in F/F patients, although without a survival difference.
What Fen Liu et al. did in 2014 was important because they tried to test that signal in a larger and more skeptical framework. In their own retrospective cohort of 164 newly diagnosed DLBCL patients treated with R-CHOP, they found no association between FCGR3A genotype and overall response rate or complete response rate. They then added a meta-analysis totaling 731 cases, and again found no significant association between the polymorphism and response in the main genetic models. Their conclusion was blunt: no clear relationship between FCGR3A 158V/F and response to frontline R-CHOP.
So in practical terms, the 2014 paper says: the early positive story did not hold up well when examined in a broader dataset. That does not mean the biology was foolish. It means the biology was probably too simple for the clinical setting. In frontline DLBCL, patients are not receiving rituximab alone; they are getting R-CHOP, where chemotherapy contributes heavily to outcome. Once you mix rituximab with multi-agent chemo, any modest Fc-receptor effect may be diluted, masked, or swamped by other determinants of response such as tumor biology, IPI risk, cell-of-origin subtype, dose intensity, and host factors. The paper itself notes that prior studies had been conflicting and that the earlier positive Asian studies were relatively small.
There is one nuance. Fen Liu et al. did report that the F/F genotype was associated with shorter progression-free survival, with borderline significance overall (p=0.05) and significance in the non-GC subtype (p=0.04), but they found no overall survival association. That kind of isolated secondary signal is interesting biologically, but it is not the same as a robust predictive biomarker for clinical use. It looks more like a hypothesis-generating observation than something you would build an LDT around.
So if you connect this back to the old PGxPredict:RITUXIMAB era, the implication is fairly clear: the later literature moved away from supporting a strong actionable claim that FCGR3A genotyping could tell clinicians who would or would not benefit from rituximab-based therapy. The 2014 paper did not merely fail to confirm the original excitement; it helped show why a plausible mechanistic marker can still be too weak, too inconsistent, or too context-dependent to survive as a commercial predictive test.
If you’d like, I can also turn this into a tight 150–200 word paragraph in your policy style, linking it explicitly to the hazards of premature predictive claims for LDTs.
