One of our collaborators Josh Mitteldorf is proposing a system for us to rapidly test not only single compounds but also how we might also test combinations of compounds and how we can measure efficacy and synergy. The MMTP is working with Josh and a number of other lifespan researchers in order to create a practical protocol for testing multiple combinations using smaller mouse groups and yet allowing us to achieve quality results. Once potential combinations are identified this then clears the way for a larger scale test on promising synergies and solutions.
We Know Nothing about Longevity Drug Interactions
Years ago, I worked for an energy conservation priest whose motto was, “anything worth doing is worth doing poorly.” We were training unemployed young people to install fuel-efficient furnaces in the homes of people who couldn’t afford heat. My boss’s point was that the new furnaces were so much more efficient than the old that even if they were installed sloppily they would save enough fuel to turn a profit.
If you focus on the big picture and get it right, the details don’t matter so much.
In tests with mice, a dozen or so different treatments have been found to lead to modest life extension. The most urgent need at present is to begin studying how these treatments work in combination. But there are too many combinations, and tests with mice are expensive. That’s why it makes sense to do a quick-and-dirty job testing all combinations at once.
I take about a dozen different pills for longevity. There is some evidence behind each of them, but what we really don’t know is how they interact. It would be nice to think that their benefits simply add, so that if one pill produces a 10% average increase in lifespan, then 10 pills increase life span 100%.
Some of them are ineffectual, of course. But for the ones that offer a benefit, most of the benefits are probably redundant. (When different treatments work via the same pathway, we can’t expect that two together work any better than either one of them separately.) A few may mutually interfere. But there also may be a few magic combinations that synergize positively. If they work via pathways that are substantially independent, we might hope that the life extension from the two together might be equal or even greater than the sum of the benefits separately.
Most of the life extension drugs that we have target a single pathway: they work through the insulin metabolism. The remainder work to suppress inflammation, or re-energize mitochondria, or lengthen telomeres, or reduce TOR signaling.
Tests in Mice and Rats
There are many ways to extend life span in worms and even in flies. Some of these have also been tested in rodents, and they don’t pass muster. I have argued that we should concentrate on mammals. Even though they are much slower and more expensive, longevity studies in mammals are a far better guide to what might work in humans. When a drug is found that extends life span in mice, there is a good chance it will also work for people (though percentage increase of our 80-year life spans is likely to be smaller than the corresponding percentage in the 2-year life span of a mouse).
Caloric restriction and exercise work consistently to increase average life span in mice. Several genetic modifications are known to work, too. The drug and supplement treatments for which there is best evidence include:
The many drugs that show promise but need further testing include
Beta Lapachone (Pao d’Arco)
Dinh lang (Policias fruticosum)
Gynostemma pentaphyllum (jiao gulan)
N-Acetyl Cysteine (NAC)
Oxytocin (not oral)
NR and NAD precursors
(Most of these were discussed briefly in a column I posted in September, and in other past columns.)
Almost no work has been done with combinations of longevity treatments. In 2013, Steve Spindler’s lab published a study based on eight different commercial formulas of vitamins and supplements. Their data were beautiful–and the survival curves for each of the eight fell exactly along the survival curve of the control group. I have heard that the NIA’s Interventions Testing Program (ITP) has tested rapamycin in combination with metformin, with successful results (to be published next year).
In a rational world, some of the billions of dollars that go into “me too” drug development and chemotherapy trials by Big Pharma would be diverted to test all of the above compounds, alone and in combination. But in the branch of the multiverse where you and I live, this will not happen in 2016. Hence “quick and dirty” (= cheap) alternatives look attractive.
The plan is to screen for combinations of drugs that offer dramatic life extension in mice, using the minimum number of mice to test the maximum number of combinations. Standard practice is to use 30-80 mice for each test in order to get a clean survival curve. The innovation I am offering is to use just a few mice for each combination of treatments so that more combinations can be tested, albeit less precisely. How many mice do we really need to be reasonably sure of not missing an outstanding combination of treatments?
I have been modeling the situation with computer-generated data, testing different statistical methods to see which works best, and how many mice are needed in order to be reasonably certain of not missing a great combination. My definition of a great combination is that it extends life span in excess of 50%. The test I propose will not be capable of distinguishing “which is better” among the rank-and-file of many treatments and combinations. However, there will be enough statistical power to identify the really hot performers, which are of most interest to us.
Specifically, I have modeled experiments based on 15 different treatments. In the most practical and successful of the methods, I combine all different triples among 15 treatments (there are 455 of them, from a well-known statistics formula Combin(15,3)). I’ve assigned 3 mice to each triplet of combinations for a total of 1365 mice.
I use statistics to tease apart the effects of different treatments and different combinations. Analysis is based on multivariate regression, but since MVR works best with just 2 or 3 variables at a time, I have been experimenting with the details of an analysis program that looks at 1 to 3 variables at a time, then does a smart search for “nearby” criteria that might do a bit better.
I generate sample data for 1365 mice, based on each mouse having a randomly life span drawn from a bell-shaped curve. The center of the bell-shaped curve depends on what treatments the mouse is getting. (This is the “right answer” that the analysis is trying to find.) The width of the bell-shaped curve is between 20 and 25% of the average, because this is the scatter that the better mouse laboratories find in their life span data.
To generate the means, I assume that each treatment offers some random amount of life extension, also drawn from a bell-shaped curve. I assume that the treatments interact in pairs and that the interactions are mostly destructive, but some of the treatment pairs interact synergistically.
For example, if a mouse is receiving treatments A, B and C, then I assume its mean predicted life span is the sum of A and B and C separately plus the three interactions (A,B), (B,C) and (A,C). (To simplify, I have assumed there is no separate term for a purely three-way interaction of (A,B,C).) This is the mean life span for that one mouse, and that mouse is assigned a life span that is a random number centered on that mean.
Since we are most interested in combinations that yield large benefit, I have adjusted the parameters so there is always at least one triple combination that (on average) has benefit of >50% life span extension.
About 40% of the time, I hit the nail on the head and identify the best triple and the best pair.
About 85% of the time, the best triple is among the top 3 generated by my analysis
About 95% of the time, the best triple is among the top 6 generated by my analysis
I think this looks promising. I am working with Edouard Debonneuil, who will check my calculations and contribute some of his own. Edouard has more experience than I have both in the practical business of managing a lab experiment and in the practical business of finding funding and sponsors.
I believe that using about 1400 mice in an experiment lasting about 3 years, we should be able to evaluate all combinations of 15 separate life extension treatments, and narrow the field to 6 candidate triples that show offer life extension in excess of 50%, and thus show promise for further testing.
…and in the Real World
The program I have outlined could be undertaken for less than the cost of testing the 15 separate treatments using traditional methodology, and I think what we would learn from the combinations protocol could be a great deal more useful. The total cost might be $1 to $3 million, depending mostly on where the work is done.
The biggest risk is that the high-benefit “magical” synergistic combinations that this program is designed to look for simply don’t exist. If they do exist and can be found, the public health impact is likely to be enormous.
But in today’s economy, who will fund this work?
National Institute for Aging in Baltimore has the Interventions Testing Program (ITP),funded at $4.7 million. Because they have high overhead and because they fund elite institutions with American salaries and because they repeat each experiment in triplicate, they can test only 1 to 2 compounds a year. There is no activity from pharmaceutical companies, except for the few compounds that can be patented. Some private foundations and crowd-funding groups have stepped forward to try to fill the void. The Glenn Foundation has cut back. The SENS Foundation is spread pretty thin. I’ve recently connected with the Major Mouse Testing Program(MMTP) of the International Longevity Alliance.
Source: The Aging Matters blog here.