Chronic myeloid leukemia (CML) is caused by ionizing radiation, such as that generated by the A-bombings of Hiroshima and Nagasaki in 1945. In the late 1990s, as a post doc at the Medical University of South Carolina, my advisor David G. Hoel asked me to apply biological understanding to mathematical models of cancer risks. The models were to be fitted to A-bomb survivor data, wherein signals are mostly at high doses, and used to predict risks at very low doses that are of interest to public health. To accomplish this, I had to focus on a cancer that was well understood and also common enough that risks of it could be detected in A-bomb survivors.  CML had this balance.

A striking feature of CML was that, although it was well understood, it was still poorly controlled. It thus contradicted the truism that understanding yields control. This truism regained its credibility only later with the arrival of imatinib.

Another striking feature of CML was that although it was thought to be simple,  out of four cohorts  of A-bomb survivors (2 cities x 2 sexes), only one, the Hiroshima males, had onset timings of 5 to 10 years as expected based on US women irradiated for cervical cancer.  Hiroshima female cases presented clinically much later and missing Nagasaki adult cases of both sexes were either never created or they existed only subclinically.  It was thus clear to me that CML was not simple.

A big difference in Nagasaki vs Hiroshima survivors was that there were 44 vs 5 adult T-cell leukemia (ATL) cases in 1.2 vs 2.7 million person-years at risk.  As HTLV-1 is the only known cause of ATL, this observation suggests that there was a 20-fold city difference in HTLV-1 seroprevalence, e.g. 5% of Hiroshima residents vs 95% of Nagasaki residents .  HTLV-1 could thus have been the cause of  radiation-induced CML incidence differences between these two cities. 

Initially I had two ideas of how CML was perhaps not created in irradiated Nagasaki adults. In one, HTLV-1 reduced target cell numbers ~10-fold by draining down quiescent stem cell reserves. In the other, it reduced the risk per target cell ~10-fold by keeping BCR and ABL1 further apart. If either of these ideas were true, however, one would expect age-induced CML incidence rates to  be reduced ~10-fold by HTLV-1, which was not observed. This reasoning convinced me that CML was created in Nagasaki, but that it was kept subclinical by immunity, as in cases of Hiroshima females with delayed onsets.  This view had practical implications, as understanding CML suppression in Nagasaki could have led to a vaccine that is used after diagnoses, a bit like using rabies vaccines after evidence of an infection.

In 2001, after the news came out that imatinib was effective against CML, I left the field. Eventually, however, as funding shifted toward immuno-oncology, I returned, as I sensed immunity was involved in the CML mysteries of Hiroshima and Nagasaki.  I started with a letter to remind people of these forgotten mysteries.  My interest in the COVID pandemic then led me to learn that the flu epidemics of 1918 and 1968 became the two main flu strains still with us today. 

Then it clicked! The four Hiroshima female radiation-induced CML onsets in 1969 to 1974 correlated perfectly with the flu of 1968 entering Japan in 1969.  Exposures to this flu strain could have weakened Hiroshima female immuno-control of subclinical CML and thus caused their releases in a cluster until all were released, at least all that could be released.

Things that still needed to be explained were Nagasaki subclinical cases existing in both sexes, their creations at lower doses, and their robustness across flu infections.  Suppression at lower doses could be explained by HTLV-1 expanding CD4-positive cell numbers, if such cells were also the target cells of radiation-induced BCR::ABL1 and the resulting clone acted as a vaccine. Suppression in both sexes that is robust across flu infections could be explained by per cell efficacy of the vaccine being stronger, as these cells had not only BCR::ABL1 junction peptide expression but also expression of HTLV-1 genes in the same cells. This explanation suggests a CML  treatment strategy that involves extracting CD4-positive cells from a patient, engineering them to express both the BCR::ABL1 junction peptide and a yet-to-be-identified antigenic peptide of HTLV-1, and expanding them ex vivo before injecting them back into the patient.  CML would then be immuno-controlled robustly in both sexes, as it was in Nagasaki. 

I pieced these ideas together in my analysis of a mathematical model of CML treatment-free remission (TFR).  The model yielded three classes of patients. My analysis of it further proposed that they corresponded to patients who were Hiroshima-male-like (not capable of a TFR), Hiroshima-female-like (capable of TFRs easily lost across immune system suppressions), or Nagasaki-like (capable of TFRs that are robust across immune suppressions).  The model used was a system of nonlinear ordinary differential equations that takes advanced training in math to understand. In our current Letter to Leukemia we present an energy landscape view of the situation that is simpler to grasp.

On the door of the autopsy room in which my father worked, the sign read mortui vivos docent, which is Latin for the dead teach the living.  This statement also applies to my work as an epidemiologist. Two bombings of two cities produced two mysteries. It is painful to think about what was done, but if we ignore data generated by these acts, we may miss the guidance it can provide toward a CD4-positive cell solution to CML.