In the spring of 2015, a couple of researchers at the recently launched MIT Center for Microbiome Informatics & Therapeutics (CMIT), including the Center’s co-director, Prof. Eric Alm, converged on a study design to better understand the impact of individual dietary micronutrients on gut microbiome composition.

The main difficulty we faced was finding a way to measure the effect of these micronutrients against a diet background that was identical in all study participants. After several iterations and discussions with nutritionists and other experts, we converged on using EnsureTM, a liquid nutritional meal replacement used in clinical contexts, as the sole source of nutrition during the study period. This would guarantee that the chemical composition of every participants’ diet was identical, and addressed one of the main concerns with previous microbiome-related diet studies.

The study design was clean. We were confident it would allow us to contribute to a number of important discussions in the field. By the summer of 2015, it looked like this study was going to happen. We had submitted the study protocol to the MIT IRB (review board), who had come back with comments that made it seem like the protocol would be accepted if we addressed several minor concerns. A couple of us decided to test the dietary regimen on ourselves, and at that moment, reality sunk in.

Within a couple of days of eating nothing but Ensure, we felt depressed, lethargic, and desperately reached for alternate sources of pleasure, now that we had removed food from the table. It is harder than it sounds: beyond the importance of food as a reward mechanism in our daily routines, there are countless social opportunities in our day-to-day lives where one consumes chemicals other than water, and which one now has to choose to avoid or suffer through without participating. A week seems like a short period of time, but unless you isolate yourself from others completely, there will be several occasions where you will be forced to weigh the benefits gained from following the study protocol until the end against your sense of self-pity and entitlement to enjoy life.

How we were going to convince 50-some healthy people to go through this very uncomfortable experience, in the name of science, was far from clear.

At the same time, several students from the Harvard-MIT Health Sciences and Technology MD-PhD program had decided to tackle the same question and had converged on an extremely similar study design. They had formulated a plan to form the HST Microbiome Consortium, and conduct the study as a class on the side of their MD program requirements. They reached out to Prof. Alm for advice, who immediately saw an opportunity for synergy. We quickly decided to join forces, combining the scientific expertise, infrastructure and access to funding at CMIT with the HST student body’s cohort, clinical vision and motivation to turn difficult projects into reality.

This was no simple feat. We had to motivate and enroll over 50 participants, randomize them to particular micronutrient spike-ins, provide them with ten stool sampling kits each, supply several pick-up points where they could obtain as much Ensure as they required for the duration of the study, and organize locations for dropping off samples that would be picked up on a daily basis.

It wasn’t long before a number of us had rented a truck and were making trips to Costco to buy all of their Ensure. It took all the available Ensure at Costco, several CVS Pharmacies, and a shipment from Amazon to reach a supply size we were comfortable with.

We then hosted an info session for interested participants, where we ran through the details of the study before obtaining their consent on an individual basis. Demonstrations of how to effectively use a stool collection kit were provided, much to everyone’s amusement. Within days, a large fraction of the cohort (over 30 individuals), populated mostly by volunteers from the HST MD-PhD program, all underwent the study during the same week. The HST lounge and Harvard Medical School classrooms became littered with empty bottles of Ensure; sample collection boxes were filled to the brim on a daily basis.

We worked around the clock to process all samples within 48 hours of receipt, treating them effectively as a single batch and allowing us to send them off for sequencing as quickly as possible.

All in all, over 200 samples were processed in a single week, and we emerged without any major incidents or mishaps. Additional enrollments to increase statistical power trickled in over the following couple of months, but the biggest lift was behind us.

Clinical studies of this sort usually roll out in three main chapters: deciding on a study design and obtaining IRB approval; enrolling patients/participants and executing the study; and performing data analysis for publication.

The second chapter was undoubtedly the most intimidating barrier to this study’s success, and we had in large part disposed of it in under two weeks. For all the complications of the downstream scientific analysis of the samples, even a layperson could appreciate that convincing dozens of healthy people to eat nothing but milkshakes and water for a week and collect daily stool samples throughout was non-trivial. We knew at this stage that we had pulled off quite a unique feat, and through collaboration had experienced something we would remember with pride for the rest of our lives.

For more on the study, including its findings, click here.

Read the resulting paper here.