A Light-hearted GREETING CARD
Good morning, Australia,
Left-e here in the USA wishing all of you a happy holiday season. We're getting some unexpected snowy weather to set the tone for our celebrations. It's so traditional! Fireplaces are being readied, gifts carefully chosen, mistletoe hung... there are colorful wrappings, children delighting, sugarplums dancing... greeting cards, carolers (with masks), tree trimming, jingle bells and cheerful exchanges.
Families are defying lockdowns to gather. After awhile some adult drinks are poured.
Yes, there's another side to all this. In short, it's a time of stress for many of us with relatives that are - let's admit it - not quite ready for the Hallmark Hall of Fame. There's Aunt Martha's bronchitis that resonates off key in the age of Covid and little Billy needs a lot of Kleenex. Your brother-in-law the investing guru is bragging about how he bought tons of Crispr at 47... cents mind you!... and now it's up to... It's the time of year not to think of commercial matters, but it all bolsters that nagging doubt: Is the goose getting cooked in my oven... or in my portfolio? And for those of us who like to post we have to battle the poco loco Yahoo censor and the Hot Copper text editor that converts plain English into a Gemoltenglobofrazzleberryjelliedgemisch. I've always felt the best gifts are hand-made, and that includes those made with a keyboard. Posting comes with the honor of garnering compliments like "clown" and "fool". Seldom any censorship there! Well, time and trials will tell. Meanwhile, it's the time of year to use our imaginations, a bit like children. That's what I plan to do with this overdue message.
The market handled the CHF results pretty well, but has had trouble digesting the ARDS report. So, let's focus on that a little, appropriate for this thread that I started a few months back. The dust has settled a bit on the decision to stop enrollment, but traders remain on edge, and much has been written by the great posters on this board with insights deeper than mine. If I'm repeating some points made by others, that's not intentional.
I offer my opinion for what it's worth and I'll start with two quotes to continue in the light-hearted spirit of the season:
- Ebenezer Scrooge, upon hearing the recommendation of the DSMB, Act 1 of
A Christmas Carol- "We're all gonna die. We are ALL going to die. There's no escaping it.
- Sigmund "Siggy" Marvin, the 10-year old son of a psychiatrist in the movie, What About Bob?
In the analysis of the C-19 phase 3 ARDS results, three things have not gotten so much attention, what I label the "Cohort Factor", the "Acme Life-Preserver Effect" and the "von Hindenburg Test for Hydrogen Effect". All had a role in how the trial played out in my opinion, and how we should interpret it.
The Cohort Factor: We are told the DSMB made its recommendation after analyzing the first 180 patients who reached the primary end-point, namely all-cause mortality at 30 days after enrollment. The recommendation was made after three steps. Let's imagine what they might have looked like. First, 90 patients were enrolled from May through July following the 12-patient EAP that had been done in April. Aside from safety, the DSMB looked at only one thing: 30 days after enrollment for any given patient, was that patient alive or dead? We all heard the conference call. SI reported that an efficacy signal was seen. He didn't tell us how strong it was, but for the sake of this example let's imagine it was a 50% reduction in mortality in the rem-L arm compared to placebo. Promising but not overwhelming... but certainly not futile either. A recommendation to continue the trial was announced. An additional 45 patients were recruited during August and September and in early November the DSMB met for a second time. They looked only at deaths in the second cohort of 45. We're told no further information was unveiled from the first cohort, so set it aside. Let's imagine at this second read-out a mortality reduction of 30% was seen in the 45-patient cohort. Still a positive signal but now below the 43% target of the study. Hmmm. Maybe a statistical fluke due to the smaller cohort size. But certainly enough to continue the trial. So a 3rd group of 45 was enrolled in October and November. Again, speculation on my part, but let's imagine the reduction in mortality of the third group came in at 10%. The shocker. Now we have 3 data points with a 50% reduction in mortality, then 30%, then 10%. Throw in the EAP that without placebo control had an imputed 80% reduction and there's a 4th data point of sorts confirming an inexorable decline (we can imagine) away from the endpoint target of a 43% reduction. Clearly over time the medicine is becoming less effective for these C-19 ARDS patients. The DSMB recommends no further enrollments due to unlikely probability of reversing this trend down. What's going on? Perhaps the medicine "doesn't really work". Is it "batch variation"? Many are already labelling it a "failure". We see gif images of Mesoplanets exploding. Management has been fustigated. Message board focus is shifting to CHF,LBP and other matters. MIS in children, pulmonary fibrosis and "long-hauler" syndrome seem forgotten. Novartis is reportedly wavering. The market has given us a lump of coal, and the analysts are saying to check back in 3 to 5 years.
"Bah, Humbug", says Ebenezer Scrooge, who disagrees with this funereal assessment. I believe he was a lefty judging from the hand he used to polish his door-knocker.
Are there any other interpretations besides failure of the medication? No way to know for sure without more information. But perhaps the illness somehow "moved the goalposts" during the 6-month enrollment period. Without getting into complex statistical analysis, let's imagine some ways that just MIGHT have happened.
The Acme LIfe-Preserver Effect: Imagine you work for the Acme Life-Preserver company, known for those handy vests stowed under your airline seat. Your boss wants to expand the business, and you're charged with collecting data on the benefits of the product. In May you head off to the airport and begin interviewing incoming passengers. After numerous interviews and 6 months of hard work you turn in your report. "Sorry, boss. Our life preservers didn't save one single life." The report makes it to the financial news. The stock of the Acme Life-Preserver Company, previously floating high, sinks to new lows. Fortunately, the CEO of the company, Sigmund Incognito, notices a technical glitch. Instead of the airport, he orders you down to the wharf. There you meet a banana boat that has just plucked passengers out of the ocean and you begin to collect statistics:
"Marvelous product, saved MY life, it did."
"Thank God for the whistle."
"The mate in 14B blew his up before exiting the aircraft. Poor soul."
You prepare a new report. Acme gets a big contract with Qantas and "SI" give you a nice year-end bonus. Notice that nothing about the life-preserver changed. Exact same product in both reports. In the first, it's a total failure. In the second a resounding success. "Hey, wait a minute" says Tom, a top-notch Australian auditor who posts prolifically on Yahoo. "How does a mistaken analysis like that first report warrant a bonus? Nobody could make a mistake that obvious, that stupid... could they?"Uhh... no, nobody could make a mistake like that, except... I'll just point out that if we're going to test whether a life vest - or a medicine - can save lives, we must find a population where lives are being lost. As standard of care for ventilator patients improves, the potential is there to gradually convert what had been a population of "banana boat" patients into "airline" patients. That has nothing to do with rem-L's effectiveness. Notice that if all patients placed on a ventilator in the placebo group "linger just a little longer" only to die on let's say Day 31, we will have a 0% mortality read-out in placebo patients. Remestemcel-L can never succeed in such a study if the only measure of success is mortality. Can't improve on 0%. Patients who get rem-L might spring out of bed, leap tall buildings in a single bound or even run for public office. Such "overwhelming" results would make no difference in a strict mortality study. The study becomes much more dependent on what happens in the placebo group, not what happens in the rem-L group. Most importantly, as we stuff the ranks of our study with airline patients they take up slots that were previously occupied by banana boat patients. As banana boat patients become rarer, we would expect to see a decrease in the APPARENT effectiveness of rem-L. We all like airline patients, we're thankful when they arrive home safely, but here they crowd out the patients we really need to study. To get statistical significance we then have to enroll a lot more total patients. Our study becomes a trip to the airport instead of a trip to the wharf. Rem-L is transformed into an unused life-preserver. Capable and perhaps highly efficacious, but not tested in the right setting with the right end point. Of course, there are ways around this problem, starting with a peek at the secondary endpoint results. Are patients leaping tall buildings? Or are they dying on Day 31? We just don't know yet. But we soon will.
So, nobody could make such an obvious mistake, except maybe in-VEST-ors who can't see past the headline. Are you one of them?
The von Hindenburg Test for Hydrogen: Could any other forces be at work? Yes, one that works though an opposing mechanism. When I was a chemistry student we did an experiment placing small strips of pure zinc metal in a test tube and then adding a few drops of hydrochloric acid, resulting in a reaction that gave off hydrogen gas. Some of us experimented with additional drops and additional reagents that for the sake of your children who might read this I won't divulge. The goal was to "speed up" the reaction. When the teacher's back was turned the mouth of one test tube was "accidentally" passed near a lit Bunsen burner. The resulting "PooooF", well let's just say the size of it erased many high school records. Yes, we were idiots, and I'm pleased to have survived intact to write about it years later. Now, imagine you work in a large city hospital testing remestemcel-L. To keep the math simple imagine 100 patients receive rem-L and 100 receive placebo. Imagine 50 die before Day 30 in the placebo group and 25 in the rem-L group. That's a 50% decrease in mortality. The trial is continued and a second cohort is recruited. This time the hospital is over-crowded so the city moves 40 of the 200 new patients into a hospital dirigible that floats above the city. 20 rem-L patients and 20 placebo patients. To keep it simple let's imagine that only patients who would have otherwise survived are moved onto the dirigible. To save money the mayor orders it inflated with hydrogen (sort of what actually happened back then, they ran out of helium). Disaster again ensues, "Oh the humanity", sad. Let's look at the results now. 50+20 = 70 patients died in the placebo group and 25+20 = 45 patients died in the rem-L group. That's a 36% reduction in mortality. In the first cohort we had a 50% reduction in mortality, now a 36% reduction. Nothing has changed about our medication, but the first result is above the target 43% reduction and the second result is below the target value. Why is this example relevant?
We can all realize that getting old is somewhat equivalent to buying a ticket on the doomed Hindenburg. "Siggy" Marvin expresses it well with his coming-of-age cry of angst. All things equal older people are more likely to die from non-ARDS problems like massive pulmonary embolus, massive stroke, massive heart attack, massive internal bleeding, massive organ failure, etc. Normally we would expect this "all-cause" mortality to decrease in patients getting better on treatment - and that's accounted for in the reduction from 50 to 25 deaths. But advanced age distributes a higher mortality risk across all patients including patients in the rem-L arm. An additional 20 deaths in each arm in this example. Nothing has changed about the medicine or its effectiveness, but statistically the "Hindenburg effect" works against proving rem-L efficacy, whether 20 patients are impacted or just 1. A tip of the hat to the great poster @
Other Perspective who pointed this out beforehand. We've been told that ventilator patients in the rem-L trial have indeed been getting older as younger patients now recover with less invasive treatments such as high flow nasal oxygen. We also learned today that 3 trials looking at FULL anti-coagulation of C-19 patients were halted due to safety concerns... it means efforts to prevent massive pulmonary embolus and stroke that I just mentioned are facing stiff competition from massive bleeding, a potential side-effect of anti-coagulation. All these complications can contribute to the Hindenburg effect. C-19 is a complex disease with much yet to be learned.... and the "standard of care" will continue to shift.
So, the Life-Preserver effect and the Hindenburg effect could have applied statistical forces to our study that worked against rem-L. Hard to know just how important a role they played, let's not exaggerate. But until we have more information on secondary endpoints, we MAY be like the Acme employee going to the airport to get statistics. In retrospect, 30-day all cause mortality may not have been the best choice in this ARDS population. It is the gold standard for judging a medication - in settings where lives can be saved. That appeared to be the case back in April when the EAP gave great hope in the early throes of the pandemic. Subsenquently, SoC improvements appear to have shifted the power of the study. Should we hold it against management? Or is management a few steps ahead of us? Just recall the timing of the Novartis deal... Personally, I'm waiting for the readout on the secondary endpoints, which could change all perspective on this promising medicine - just as the secondary endpoints are saving the day for rexlemestocel-L in CHF.
It brings me to some final points on the decision to stop enrollment in the rem-L trial early. It's interesting to compare two other trials we heard news on recently. The first is the "Ruxcovid" trial run by Novartis and reported on December 14. That trial enrolled 432 patients randomized 2 to 1 to receive ruxolitinib (Jakafi) vs placebo. These were patients who had severe C-19 pneumonia, but were not sick enough to require intubation (or gain entry to the rem-L trial). To me it's interesting that this trial was allowed to go to full enrollment despite no evidence of efficacy in the primary endpoint or any secondary endpoint. Ruxolitinib is demonstrably more toxic than rem-L. It's a small molecule JAK inhibitor acting inside cells to gum up enzymatic reactions. In other words it's a "poison". Take enough of it for long enough and can get bone marrow depression manifested by serious bleeding, serious infections, and serious anemia. Among other problems. Yet, nearly 300 patients were allowed to receive treatment with this molecule in a study that showed zero benefit. As an aside, per Dr. Grossman this is the medicine that rem-L would be tested against if a double-blinded trial to treat adult GvHD is done. Ruxolitinib's failure also suggests that care has to be taken using anti-cytokine storm meds too early (yes I'm familiar with the baricitinib results). Using a fire extinguisher the day before a fire doesn't accomplish much. Secondary endpoint data are needed! The second trial is the DREAM-HF trial that enrolled over 500 patients. That trial involved giving injections directly into cardiac muscle. The study showed no benefit for the primary end-point of hospitalizations, but jaw-dropping efficacy on secondary end points.
So, two studies with failed primary end-points that nevertheless went on to full enrollment. Compare rem-L which has shown no serious toxicities, does not involve an injection into a major organ, showed a primary efficacy signal in its EAP and early cohort, and for which I've just given possible reasons for study futility unrelated to the efficacy of rem-L. Yet its enrollment is being stopped 77 patients shy of full enrollment. The primary role of a DSMB is to protect patient safety, to call off a therapeutic boxing match if one side, rem-L vs placebo, is getting pummeled to death. That was not the case here; I believe other forces are at play.
It's possible NIH rules for financing a study require a halt if the DSMB makes any recommendation of likely futility. However the decision to halt further enrollment in the trial came about, my working hypothesis is that Mesoblast in close consultation with Novartis wanted the trial to stop in order to regroup and most importantly to save time getting to the blinded secondary endpoint data. We need to see the full data set to know exactly what happened here. I'm not willing to give up on rem-L just yet. I'm not going to make judgements based on analysis done "at the airport". I want to see full data from down at the wharf. I refuse to believe Novartis entered into this deal without having done their own extensive research of the product including patent and manufacturing/scale up assessment, lab analysis and possibly their own animal testing. I suspect they're interested in more than just ARDS and pulmonary fibrosis, might see a tie-in with their CAR-T program, might have an interest in the cardiac product. I commend management for having organized the C-19 ARDS program on short notice, partnering with the NIH, moving briskly to a phase 3, taking the ARDS lead in the industry, and signing a big pharma partner as a result. Could there be some snags along the way, some missteps on endpoint choice as the illness evolves? Of course. Should SI be pilloried as a result, or demoted, or even fired as called for by some posters? Of course not. Should investors give up hope? For each to decide.
Fortune favors the bold. We will get through this... as long as Santa can squeeze a life vest over his rotund self, jettison that bag of coal, and make it to the nearest wharf. Maybe he'll need an infusion of rem-L to treat his ARDS. It can get right frigid in northern seas this time of year.
Wishing all of you happy investing on into the New Year 2021, Left-e