A common experience and frustration in the field of condensed matter physics is variability among samples. Every device shows some differences due to variations in material quality, fabrication, etc., although generally similar phenomena can usually be seen despite these uncontrollable factors. This problem is exponentially worse in biological research, especially cancer research. Not only does every patient have different environmental and genetic factors, but every tumor is heterogeneous. The properties of cancer cells can vary throughout a tumor, meaning that treatment might be effective in some parts and ineffective in others. [1] The way to prove effectiveness despite this literally uncountable number of variables is through well-designed experiments, the gold standard being phase III randomized controlled clinical trials. Yet important fundamental questions, such as “Why do some patients respond to immunotherapy and others do not?” remain somewhat mysterious. Moreover, running these trials takes many years and hundreds of lives, limiting the number of variables investigators can test.

After receiving my diagnosis and establishing my care team, I moved on to endless research into the different aspects of human life that may be modestly modified in an attempt to amplify treatment (knock on wood). My evidentiary bar will be much lower than the one doctors require for guidance since phase III trials on many of these “interventions” have not been performed. So, my strategy has been to look for interventions with three properties: some plausible biological mechanism, at least a little human evidence, and a very low cost if the hypothesis turns out to be wrong (knock on wood). I am not replacing proven treatment with alternative medicine. I am looking for marginal gains around the edges of proven treatment. And I hope this post can possibly serve as a data point for those who come after.

The primary treatment

First, the most important part of this post: I am enrolled at Memorial Sloan Kettering in phase II trial MSK 26-102 / NCT07631000, studying perioperative cadonilimab plus FLOT for resectable, locally advanced gastric and gastroesophageal-junction adenocarcinoma. [2]

Cadonilimab is an unusual immunotherapy. It is a bispecific antibody that simultaneously targets PD-1 and CTLA-4, two immune checkpoints cancer can exploit to suppress T-cell activity. [3] FLOT is the decidedly less elegant acronym for four chemotherapy drugs: 5-fluorouracil, leucovorin, oxaliplatin, and docetaxel. The trial begins with an induction dose of cadonilimab alone, which is an interesting distinction from the standard durvalumab + FLOT procedure. Two weeks later, cadonilimab is combined with FLOT every two weeks for four preoperative treatments. Surgery follows four to eight weeks later, knock on wood. After surgery, there are another four cadonilimab-plus-FLOT treatments, followed by ten 28-day cycles of cadonilimab maintenance given twice per cycle. [2]

Everything below is secondary to and concurrent with that treatment.

Intervention I: Walking as medicine

The intervention with stronger evidence than the rest of my little experiments is also the simplest: brisk walking.

In 2025, the New England Journal of Medicine published the phase III CHALLENGE trial, which found that structured exercise resulted in significantly better disease-free survival: 80.3% versus 73.9% at five years, corresponding to a hazard ratio of 0.72 for recurrence, a new primary cancer, or death. Participants were encouraged to increase aerobic exercise by roughly 10 MET-hours per week and sustain it. That is approximately 150 minutes of moderate-intensity aerobic activity per week; a reasonable approximation is 45–60 minutes of brisk walking three or four times per week. This intervention is easy to incorporate, good for overall health, and is now potentially supported by phase III data (albeit in a totally different situation from the one I am in). [4]

Intriguingly, there was no significant difference in deaths without recurrence or without a second primary cancer (1.3% versus 1.8%), suggesting that the main benefit of exercise came from improved cancer outcomes. That is, this study, in agreement with previous observational studies, suggests that exercise can have anticancer effects, beyond improvements to general health. [4]

My rough target will be the CHALLENGE dose—about 45–60 minutes, three to four times per week.

Intervention II: Immunotherapy before lunch

The immune system has a clock. T-cell trafficking, cytokine production, checkpoint expression, and other components of immune function oscillate with circadian rhythms. That raises a straightforward question: does immunotherapy work differently depending on what time you receive it? A surprisingly large number of studies suggest it might. For instance, a 2026 meta-analysis combining 29 studies involving 6,129 patients found that immune-checkpoint treatment given earlier in the day was associated with better progression-free survival (PFS) and overall survival (OS). Intriguingly for me, the gastric-cancer subset also favored earlier treatment, although the esophageal-cancer subgroup did not show a statistically significant OS benefit. [5]

However, as with all the rest of the interventions on this list, there is an enormous asterisk. Twenty-seven of those 29 studies were retrospective. Scheduling is not randomized in the real world as healthier patients and, importantly, those living close to the treatment center may be more likely to get morning appointments. Patients receiving many cycles create their own statistical biases. Different studies define “early” using different cutoffs. A recent analysis highlighted cycle-number bias, immortal-time bias, scheduling drift, and other confounders as unresolved problems, and a secondary analysis of the randomized NCI/SWOG S1404 melanoma trial from Fred Hutch found no detectable time-of-infusion effect after accounting for these issues. [5,6]

This field, chronotherapy, is thus extremely controversial. Given the reported large signals and the simple, noninvasive nature of the intervention, it would be natural to try randomized clinical trials, instead of retrospective ones, to find the ground truth. Intriguingly, the randomized phase III trial included in that 2026 meta-analysis initially reported a spectacular advantage for earlier immunochemotherapy in lung cancer. However, it was subsequently retracted by Nature Medicine, leaving that meta-analysis without a valid randomized trial supporting the hypothesis. [5,7]

I only learned about this intervention by reading the reference list of an unrelated paper (see next intervention) after my induction dose of cadonilimab. This may have resulted in a mistake (God forbid) in my marginal-gains ethos: my first immunotherapy dose came later in the day (around 5 p.m.). Limited clinical evidence and work in mouse models suggest that the first dose may play an outsized role in determining the immune response. A preclinical study found strong time-of-day effects on checkpoint blockade in mice, while a small cohort of 84 patients with hepatocellular carcinoma found that a morning first dose was associated with better PFS and response while the timing of subsequent infusions was not associated with outcomes. [8,9] In other words, the first encounter between the immune system, tumor, and checkpoint inhibitor may establish an immune program that is disproportionately important. Evidence around this extraordinary claim remains mixed. A large VA target-trial emulation found a modest survival advantage when the first three ICI infusions were given before noon, but did not find a strong first-dose-only effect. Moreover, a 349-patient London NSCLC cohort found no significant difference based on the timing of the first dose alone, although receiving more than 50% of doses in the morning was associated with better OS. [10,11]

Therefore, going forward, the decision is easy. I immediately switched all of my treatments to the morning in case there is any benefit. I would strongly suggest those who come after do this for the first dose as well.

Intervention III: A brief fast before immunotherapy

I learned about chronotherapy in the references of a totally unrelated, recent Cell Metabolism paper titled “16-h fasting optimizes cancer immunotherapy in mice and humans.” The authors found that a short fast altered nutrient competition inside the tumor microenvironment. In their experiments, fasting increased intratumoral isoleucine availability, which supported metabolic and epigenetic programs in cytotoxic CD8 T cells. In mice, this improved checkpoint-blockade activity. In a small prospective cohort of colorectal-cancer patients receiving neoadjuvant immunotherapy, fasting was also associated with greater CD8 T-cell clonal expansion and cytotoxic programs. The human component was small, preliminary, and focused heavily on biological endpoints rather than the sort of clinical endpoint I actually care about—namely, not dying. [12]

Still, a short overnight fast is relatively easy for me. So, I chose to follow the protocol used in the paper: stop eating by about 8 p.m. and fast until about noon, roughly two hours after the immunotherapy infusion. [12] I am not doing multiday fasting around chemotherapy, both out of respect for the clinical-trial protocol and because there is no established evidence that prolonged fasting during a 24-hour 5-FU infusion is beneficial or safe in my setting. The shorter fast will also help avoid unwanted weight loss.

As with Intervention II, I only learned of this paper after my first immunotherapy dose. For those who come after, I would consider adopting it for the first dose as well.

Intervention IV: The 12-to-10 rule

The more general version of Intervention III.

Most days, I will try to limit eating between approximately 12 p.m. and 10 p.m. without any calorie restriction. That gives me a ten-hour eating window and roughly a fourteen-hour overnight fast. This choice was partly motivated by a small (43-patient) phase I/II study presented at ASCO involving patients with advanced head-and-neck cancer receiving pembrolizumab. The study reported no intervention-related weight loss and substantially higher disease-control rates in the time-restricted-eating group, along with changes in microbial metabolites and insulin-related signaling. [13]

Here’s the asterisk: this trial was small, nonrandomized, involved another cancer entirely, and remains an ASCO abstract and not a publication. The broader fasting literature is similarly suggestive but immature. Reviews generally find that time-restricted or intermittent fasting appears feasible in selected cancer patients, but there is nowhere near enough evidence to conclude that it improves cancer outcomes yet. ASCO’s dietary guideline explicitly states that the evidence is insufficient to recommend for or against fasting as a cancer-treatment intervention. [14]

I consider the evidence for this intervention to be quite weak. However, I also find it unlikely to have negative effects (knock on wood) so long as I maintain my weight, so I am adopting it nonetheless.

Intervention V: Eliminating cannabis

Contrary to the common perception of weed users as lazy losers, I have found cannabis quite useful for unlocking my mind and having unconventional, creative ideas for my physics research. As a result, cannabis was my occasional “drug of choice,” in contrast to alcohol (which I very rarely consumed even before the diagnosis).

In the cancer context, cannabis has real medical potential for symptom management, including relief of pain, nausea, and loss of appetite. [15] Yet one of my interventions is cutting out cannabis entirely. In addition to some work in mouse models, there are a couple of Israeli studies suggesting that cannabis use might interfere with immune-checkpoint therapy. The first looked retrospectively at 140 patients receiving nivolumab. The response rate was 37.5% without cannabis versus 15.9% with cannabis, although the study did not find a statistically significant difference in PFS or OS. [16] A later prospective observational study from the same research community followed 102 patients and did report substantially shorter time to progression and OS among cannabis users. [17] The proposed mechanism involves the broad role of the endocannabinoid system in modulating immune function, with preclinical evidence of immunosuppressive effects from cannabinoids. [18]

Those numbers certainly attract one’s attention when one happens to be receiving immunotherapy. The studies are also, scientifically speaking, sketchy as fuck, with concerning comments on PubPeer regarding their statistical analysis. More importantly, they were small, involved heterogeneous cancers and treatments, were not randomized, and have enormous potential for confounding factors. For instance, cannabis use may be driven by worse symptoms, more advanced disease, etc., that independently predict poorer survival. Moreover, subsequent works have made the picture considerably murkier. A 2025 systematic review found little convincing evidence that cannabis produces the clinically meaningful immune suppression one would expect if it were sabotaging checkpoint inhibitors. [18] A pooled analysis of 684 participants from four Canadian immunotherapy trials likewise found no significant detrimental association between cannabinoid use and conventional PFS, OS, response, or immune toxicity. [19] Preclinical models have similarly produced mixed results. [18]

So: cannabis has not been proven to reduce immunotherapy effectiveness. Still, the calculation for me was easy. There is a biologically plausible downside, however uncertain, so I stopped entirely for the duration of my treatment. As I said before, I would give up any vice for even a one-percent increase in survival (knock on wood).

Intervention VI: (somewhat) Reduced sugar

Cancer cells famously love glucose. One of the defining metabolic features of many tumors is increased glucose uptake and glycolysis, a fact exploited by PET scans. The tracer FDG is a radioactive glucose analogue, and tumors often light up because they take up and metabolize glucose more rapidly than surrounding tissue. [20] It is therefore tempting to reason: Cancer likes glucose → stop eating sugar → starve cancer. Unfortunately, human metabolism is not that easy. Our bodies maintain blood glucose, and cancer is also metabolically adaptable. The National Cancer Institute notes that, although cancer cells generally consume more glucose than normal cells, there is no evidence that eating sugar makes an existing cancer grow faster or that eliminating sugar makes it shrink. [21]

So I have cut back substantially on added sugar, but I am not trying to entirely eliminate carbohydrates or achieve nutritional ketosis. For me, this is another asymmetric bet. Large amounts of nutritionally useless added sugar do not offer much benefit. Reducing them is easy. But maintaining weight does offer a benefit. If calories become difficult to obtain and the choice is between some theoretically immaculate metabolic diet and a milkshake I can actually swallow, I am drinking the milkshake or eating the occasional dessert to keep life worth living.

An ending note

Exercise, morning appointments, eating windows, cannabis abstinence, and dietary choices are experiments around the margins. I do not know whether any of them will add a single day to my life. It is entirely possible that they will not (God forbid). I am one person running an uncontrolled experiment with an n of one. Nothing that happens to me will prove or disprove the effectiveness of any of these interventions.

But I can at least leave behind the map I used, the evidence I found interesting, the evidence I found dubious, and the mistakes I may have made along the way.

For those who come after.

References

  1. [1]National Cancer Institute. “Tumor heterogeneity.” NCI Dictionary of Cancer Terms. Accessed September 6, 2026. Link ↗
  2. [2]National Cancer Institute. “A Study of Cadonilimab Combined with Chemotherapy in Treating Patients with Gastric or Gastroesophageal Junction Cancer.” Phase II trial MSK 26-102; NCT07631000. Accessed September 6, 2026. Link ↗
  3. [3]Pang X, Huang Z, Zhong T, et al. Cadonilimab, a tetravalent PD-1/CTLA-4 bispecific antibody with trans-binding and enhanced target binding avidity. MAbs. 2023;15(1):2180794. doi:10.1080/19420862.2023.2180794. Link ↗
  4. [4]Courneya KS, Vardy JL, O’Callaghan CJ, et al.; CHALLENGE Investigators. Structured Exercise after Adjuvant Chemotherapy for Colon Cancer. N Engl J Med. 2025;393(1):13–25. doi:10.1056/NEJMoa2502760. Link ↗
  5. [5]Inoue S, Tsuboi I, Miszczyk M, et al. Time-of-Day Immunotherapy Administration and Outcomes in Advanced Cancers: A Systematic Review and Meta-Analysis. JAMA Netw Open. 2026;9(5):e2610815. doi:10.1001/jamanetworkopen.2026.10815. Link ↗
  6. [6]Othus M, Truong T-G, Sharon E, et al. Addressing Biases in Analysis of Time of Infusion: NCI/SWOG Trial S1404 Among Participants With High-Risk Resectable Melanoma Who Received Adjuvant Anti-PD-1 Therapy. JCO Oncol Pract. Published online June 23, 2026. doi:10.1200/OP-25-01413. Link ↗
  7. [7]Huang Z, Zeng L, Ruan Z, et al. Retraction Note: Time-of-day immunochemotherapy in non-small cell lung cancer: a randomized phase 3 trial. Nat Med. 2026;32:2691. doi:10.1038/s41591-026-04508-1. Link ↗
  8. [8]Lichterman JN, Srinivasan T, Wang R, et al. Timing of immune checkpoint blockade shapes anti-tumor immunity via a clock-dependent chemokine axis. bioRxiv [Preprint]. 2025. doi:10.1101/2025.11.03.686293. Link ↗
  9. [9]Li HL, Charmsaz S, Reisman BJ, et al. Time-of-day of first checkpoint inhibitor dose influences clinical outcomes and immune responses in hepatocellular carcinoma. J Immunother Cancer. 2026;14(4):e014070. doi:10.1136/jitc-2025-014070. Link ↗
  10. [10]Gonzalez RT, Datta R, Rehmani S, et al. Morning versus afternoon administration of immune checkpoint inhibitors in metastatic non-small-cell lung cancer. J Immunother Cancer. 2026;14(5):e013830. doi:10.1136/jitc-2025-013830. Link ↗
  11. [11]Gomez-Randulfe I, Pearce M, Netto D, Ward R, Califano R. Association between immunotherapy timing and efficacy in non-small cell lung cancer: a comprehensive analysis at a high-volume specialist centre. Transl Lung Cancer Res. 2025;14(1):72–80. doi:10.21037/tlcr-24-571. Link ↗
  12. [12]Chen S, Hu T, Zhu K, et al. 16-h fasting optimizes cancer immunotherapy in mice and humans. Cell Metab. 2026;38(5):944–962.e14. doi:10.1016/j.cmet.2026.01.015. Link ↗
  13. [13]Sarfraz H, Bari S, Eysha M, et al. Safety and efficacy of time restricted eating (TRE) in improving response to immunotherapy in patients with head and neck cancer (HNSCC). J Clin Oncol. 2023;41(16_suppl):6051. doi:10.1200/JCO.2023.41.16_suppl.6051. Link ↗
  14. [14]Ligibel JA, Bohlke K, May AM, et al. Exercise, Diet, and Weight Management During Cancer Treatment: ASCO Guideline. J Clin Oncol. 2022;40(22):2491–2507. doi:10.1200/JCO.22.00687. Link ↗
  15. [15]National Cancer Institute. Cannabis and Cannabinoids (PDQ®)—Patient Version. Accessed September 6, 2026. Link ↗
  16. [16]Taha T, Meiri D, Talhamy S, Wollner M, Peer A, Bar-Sela G. Cannabis Impacts Tumor Response Rate to Nivolumab in Patients with Advanced Malignancies. Oncologist. 2019;24(4):549–554. doi:10.1634/theoncologist.2018-0383. Link ↗
  17. [17]Bar-Sela G, Cohen I, Campisi-Pinto S, et al. Cannabis Consumption Used by Cancer Patients during Immunotherapy Correlates with Poor Clinical Outcome. Cancers (Basel). 2020;12(9):2447. doi:10.3390/cancers12092447. Link ↗
  18. [18]Behling-Hess C, Simonson G, Salz T, Fleege N, Zylla D. The impact of cannabis on immune checkpoint inhibitor therapy: a systematic review of immunomodulatory effects of cannabis in patients with and without cancer. Support Care Cancer. 2025;33(3):166. doi:10.1007/s00520-025-09218-x. Link ↗
  19. [19]Coschi CH, Ding K, Tong J, et al. Effects of cannabinoids on immune checkpoint inhibitor response: CCTG pooled analysis of individual patient data. Immunotherapy. 2025;17(4):257–268. doi:10.1080/1750743X.2025.2485012. Link ↗
  20. [20]National Cancer Institute. Computed Tomography (CT) Scans and Cancer Fact Sheet, section “What is combined PET/CT?” Accessed September 6, 2026. Link ↗
  21. [21]National Cancer Institute. Common Cancer Myths and Misconceptions, “Will eating sugar make my cancer worse?” Accessed September 6, 2026. Link ↗

This is a personal account of my own experience and research, not medical advice. Treatment decisions belong with qualified clinicians and, in my case, must remain consistent with the clinical-trial protocol.