China Fast-Follower Competition Reaches Clinical CRISPR

There is panic among Western biotech that Chinese competitors will eat their lunch with their capital-efficient fast-follower strategy which typically involves rapid clinical translation via investigator-initiated trials.  This issue has now reached the CRISPR space in the form of first clinical data announced by YolTech regarding a PCSK9 base editing trial for the treatment of hypercholesterolemia.

Almost identical to pioneer Verve Therapeutics which reported stellar data earlier this month (discussed here), YolTech’s YOLT-101 formulation involved a GalNAc-LNP encapsulating an adenine base editor mRNA and guide RNA targeting a splice site of the PCSK9 pre-mRNA for gene knockout.  The more detailed nature of the LNP formulation was not disclosed in the accompanying medRXiv publication.   

The trial tested 3 dose levels of YOLT-101: 0.2mg/kg (n=1), 0.4mg/kg (n=2), and 0.6mg/kg (n=3) indicating an unusually rapid move up in the dose level by international standards.  Efficacy was only reported for one subject treated with 0.4mg/kg and the three 0.6mg/kg subjects with heterozygous familial hypercholesterolemia.

Similar to Verve Therapeutics, LDL-cholesterol lowering was roughly -50% for 0.6mg/kg.  Unfortunately, the information provided did not allow for an analysis of the relationship of total dose of YOLT-101 and LDLc reduction.  On the PCSK9 front, YolTech seemingly did better than Verve Therapeutics reaching a mean of -76% versus the -60% for VERVE-102 both at 0.6mg/kg.

This, however, is where the similarities ended.  In terms of the critical safety of a potentially very widely applicable therapy, 3 out of the 6 subjects treated with YOLT-101 exhibited ‘transient elevations in ALT and AST’ that ‘almost’ returned to normal within one month.  Furthermore, 5 of 6 subjects experienced infusion-related reactions involving fever, myalgia, and vomiting. And similar to Verve’s ill-fated VERVE-101 formulation, one subject at the 0.4mg/kg dose experienced chest pain shortly after LNP infusion.

Nevertheless, the authors noted that the trial remains ongoing to ‘validate the therapeutic durability and safety profile’.  Considering the ALT/AST elevations for which more detailed values were not disclosed, it seems questionable whether this is an ethical decision.  Add to this the rapid dose escalation and selective data disclosures, it provides fodder to those criticizing China for allowing human experimentation and Big Pharma taking advantage of it by licensing therapeutic candidates built on such strategies on the cheap, not even mentioning the intellectual property issues of ‘Chinese Beam Therapeutics’

The time for genome editing is now

Back from recent CRISPR conferences in Ireland and Denmark, I wanted to share my current thinking about the present and foreseeable future of clinical genome editing.

CRISPR is a clinically mature platform

First of all, genome editing, and in particular CRISPR Therapeutics have arrived big time.  Ignore for a moment the 90%+ declines in the share prices of Intellia, Verve, and Metagenomi, that all, with the exception of perhaps Beam Therapeutics, suggest that the industry will not survive.  Focus instead on the transformational results last week by Verve Therapeutics which will shape the management of chronic metabolic and cardiovascular disease for decades to come, Beam Therapeutics last month announcing that it  for the first time fixed a genetic disease (AATD) by reversing a point mutation, or the three ongoing phase 3 programs by Intellia Therapeutics which promise patients live unintruded by their disease.

So you say that this is only half the truth and many of them will develop cancer?  With hundreds of subjects having been given CRISPR therapeutics, we still have yet to hear about such cases, including for the hematopoietic stem cell applications, probably the most stringent test in that regard given the history of retroviral gene therapy.  This is supported by ample detailed genomic analyses demonstrating that many CRISPR modalities, in particular prime editing, often do not have any detectable off-targeting (!) and therefore are safer than going out in the sun or eating grilled meat.  And don’t get me started with analysts and investors pretending that small molecule drugs, 'dirty drugs' by design, are much safer in that regard when many of them like methotrexate and hydroxyurea are known to cause cancer which barely gets a mention or is routinely accepted as part of the risk-benefit of a drug.  Similarly, the currently more widely embraced antisense and RNAi technologies exhibit considerably more off-targeting, which in theory could result in cancer development with chronic use, than the average CRISPR therapeutic candidate today. 

Prime editing is the ultimate CRISPR manifestation

Prime editing is the foreseeable endgame for genome editing involving edits up to 50 nucleotides.  This is because in addition to being able to knock out genes like (Cas9) nuclease or base editing, it allows the developer to change nucleotides at will and with incredible specificity, and for gain-of-function.  It has therefore surprised me to see that prime editing is still such a niche part in these conferences and barely used in industry and even academia where IP is less of an issue.  This can likely be explained with it requiring a good deal of mental gymnastics when devising a prime editing strategy and subsequent tinkering to optimize efficacy.  Prime editing will likely also be part of strategies for achieving even larger edits, including whole gene replacements that are often desirable.

In some ways this is reminiscent of zinc fingers which make for poor screening and quick molecular biology tools and were quickly forgotten when CRISPR came along with the promise of simple sgRNA design.  However, a panel at the Copenhagen CRISPR meeting by genome editing veterans made it clear that when optimized, zinc fingers can be every bit as powerful as CRISPR and currently have a leg up where delivery is going to be viral for some time to come.  

This is particularly true for the CNS, the fourth major tissue/organ target for genome editing applications after the liver, hematopoietic stem cell, and immune cells.  Not only is it impossible to fit in prime and base editors in single AAV vectors, expressing a CRISPR effector nuclease of bacterial origin for years may be tricky (but not unsolvable) in terms of immunogenicity.  There is also some unease about having nucleases loiter around in cells for years after their job has been done.  Self-targeting/inactivation strategies may here be called for.  Zinc fingers by being small and mimicking human proteins by contrast are ready for clinical use, a fact, as the deal flow by Sangamo Therapeutics illustrates, is increasingly being recognized by the wider industry.

Genome editing versus non-permanent modalities

Another popular argument against CRISPR Therapeutics questions why one should take a ‘risk’ with permanently altering the genome when you have oligonucleotide therapeutics options that need to be administered as infrequently as every quarter, half annually, or even annually.  

While there is some application overlap between genome editing and oligonucleotide therapeutics, especially when it comes to knocking down genes, a closer look shows that the overlap is limited.  Where there is clear genetic evidence that a gene can be disabled for life without safety risks, then genome editing is the endgame, or at least capture a good portion of the market, and transient modalities may serve to de-risk the targets first.  

When it comes to bringing pathways back into finely tuned balance and/or the disease is not chronic and likely requires only temporary treatment (e.g. some cases of hypertension), transient options are preferable.  Still in other cases such as prion or Huntington’s disease where it is not clear whether it is the overall level or distribution of gene knockdown that is more important for efficacy, the preferred modality has yet to emerge from clinical studies. 

Part of the Wall Street fashion show

Wall Street is a fashion show when it comes to biotech platforms.  A little less than 10 years ago, it predicted that CRISPR would kill off the RNAi Therapeutics industry. Today, Alnylam sports a $30B market cap, about 4x that of the entire CRISPR industry combined (much of it in cash).  Of course, investing in technologies ahead of their time is fatal. The electric vehicle industry is just one example.  As I have laid out above, genome editing, however, is now and here to stay.  Having the potential to save the healthcare system money (e.g. $2M for a one-and-done HAE/ATTR treatment as opposed to $500k annually) or to lead to drastically improved outcomes (as can be predicted from chronic LDLc lowering) got to be embraced by patients, their caregivers, docs, regulators, and insurers alike.  

It's been a rough couple of years investing in development-stage, cutting-edge biotechs and current politics does not seem helpful, but if this is when it is darkest, then generational wealth awaits the survivors with the guts to hold on during a volatile comeback.  One of my two conviction bets that I look up to in this regard is therefore CRISPR therapeutic biotech Verve Therapeutics.  Uniqure, which works in the similarly unloved gene therapy arena and has a promising DNA-directed RNAi candidate for Huntington’s disease, is the other in case you were wondering.  

Verve Therapeutics Nails Cardiovascular Disease CRISPR Study

Patients do not benefit from drugs they do not take.  This is especially true in the cardiovascular disease space aimed at lowering atherogenic LDL-cholesterol where the majority of patients starting on oral options like statins do not take their pills after just one year.  This is also true for once every 2 to 4 weeks next-generation PCSK9 antibodies and even semiannual PCSK9 RNAi therapeutic inclisiran, though to a lesser degree in the latter case.

With this realization in mind, Verve Therapeutics set out to develop a PCSK9 CRISPR base editing treatment that should lower LDL-cholesterol for life by at least -40% after just a single administration of an intravenous LNP formulation.  Unfortunately, a first generation formulation, VERVE-101, had to be abandoned a year ago because of laboratory abnormalities, in particular ALT/AST elevations 5 to 10-fold above the upper limit of normal as well as a case of dangerously low platelet counts in a first clinical trial.  In addition, the intra-dose variability of the PCSK9 knockdown and LDLc lowering between subjects and the dose-responsiveness were not optimal.

Liver enzyme elevations (here ALT) with VERVE-101 in the HEART-1 study

All evidence pointed towards the LNP formulation, not the PCSK9 as the target or the base editing process, to be the culprit for the safety issues.  Verve therefore decided to replace some of the lipids in the liposomal formulation and add GalNAc sugars so as to allow the LNP to be taken up by both the LDL-receptor (via ApoE)- and ASGPR (via GalNAc).  This is helpful for two patient populations that are most in need for new treatment options and which lack LDL receptors (heFH and hoFH).  The base editor and guide RNAs were left unchanged from VERVE-101. 

Based on data from the first 14 subjects treated with VERVE-102 revealed today the theory translated perfectly into clinical practice.  At doses above 50mg of the LNP, the mean LDLc reduction was -59%, in line with the most potent PCSK9 agents (antibodies) and significantly more potent than inclisiran, especially in the heterozygous FH (heFH) population.  Moreover, there was a beautiful dose response for both PCSK9 and LDLc lowering and very little inter-patient variability.

Dose-related LDLc lowering in the HEART-2 trial with VERVE-102

Even more importantly, the safety was pristine.  There was hardly a blip with no outliers in terms of ALT/AST changes upon LNP administration, a stark difference to VERVE-101.  Similarly, no platelet changes were seen.  Only a single case of grade 2 infusion reaction was observed which rapidly resolved and does not pose an obstacle to further clinical development and commercialization.  Anybody familiar with LNP technology understands that GalNAc-LNPs are now the gold standard in the delivery of genome editing in the liver.

Verve Therapeutics is wrapping up the HEART-2 study with a final higher dose to see whether there is further LDLc lowering and then proceed to a ~60-subject phase II study aimed at locking down one of two fixed doses of VERVE-102 for the registrational phase of clinical development.

Today marks a milestone in moving genome editing to large, indeed very large patient populations. 

Disclosure: I owned some Verve Therapeutics shares going into data and doubled down on it after seeing the emerging VERVE-102 product profile.