Every August, something odd happens in our intake logs. The CD34+ counts — the progenitor cells we track as a primary quality indicator for peripheral blood donations — slide downward in a pattern that repeats so reliably it started to feel like a calendar event rather than random noise. We noticed it first in the 2020 data, dismissed it, then saw it again in 2021 and 2022, at which point it became hard to ignore. So in early 2024, we sat down with five full years of intake records, 2,847 samples collected between January 2019 and December 2023, and asked the question properly: is there a real seasonal signal in peripheral blood CD34+ counts, and if so, what is driving it? The answer turned out to be more complicated than we expected, and the implications touch on everything from donor scheduling to mobilization protocols.

What the data actually shows, month by month

Averaging CD34+ cells per microliter across all five years by calendar month produces a curve with a shape that is difficult to unsee once you have plotted it. January through March sits at a relatively stable band, with a mean of 3.41 cells per microliter and a standard deviation of 0.88. April begins a gentle decline. By July the mean drops to 2.67 cells per microliter, and August is the single lowest month across all five years at 2.51. September starts a recovery, and by November the figures are back near the January baseline. The trough is not dramatic in absolute terms — we are talking roughly a 26 percent reduction from peak to valley — but it is statistically robust. A one-way ANOVA across the 12-month bins returns F(11, 2835) = 4.87, p less than 0.001, and post-hoc Tukey tests confirm that July and August are significantly different from January, February, March, and November. The effect holds when we subset by sex, by donor age band, and by the three different mobilization protocols in use across the study period.

Building a credible dataset: how we cleaned five years of intake records

The 2,847 records come from peripheral blood samples collected after G-CSF mobilization, processed at our Bunkyo facility using a BD FACSCanto II flow cytometer with the ISHAGE gating strategy. We excluded 214 records where the mobilization course was interrupted, 38 records flagged for instrument calibration issues on the collection day, and 9 records where the donor had a concurrent acute infection confirmed by CRP above 10 mg/L. That left a working dataset of 2,586 fully usable samples. Donor demographics across the five years are reasonably stable: median age 34 years, 61 percent male, with no significant year-on-year drift in the age or sex distribution. We did not attempt to control for donor body weight in the primary analysis, though a sensitivity run weighting counts by body surface area produced a nearly identical seasonal curve, which gives us some confidence that the pattern is not an artifact of who happens to donate in summer versus winter.

Temperature, hydration, and the hematological plausibility question

The biological literature offers several candidates for a summer suppression of circulating progenitors. Heat stress has been shown in animal models to reduce bone marrow output, though the effect sizes in rodent studies are larger than what we observe here. A more grounded explanation is hemodilution. Tokyo summers are genuinely punishing — heat index regularly exceeds 35 degrees Celsius through July and August — and donors arriving for mobilization are likely consuming more fluid and sweating more heavily than in cooler months. Plasma volume expansion without a proportional rise in cell output would dilute measured counts. We attempted a rough correction using hematocrit as a proxy for hydration status. Donors with hematocrit below 40 percent on collection day, which skewed toward summer months, did show lower CD34+ counts on average (2.44 vs. 2.79 cells per microliter in donors above that threshold). But even after binning by hematocrit quartile, the seasonal dip persists, suggesting hemodilution alone cannot carry the full explanation.

The mobilization protocol confound we almost missed

Here is where it gets methodologically uncomfortable. In June 2021, we shifted a portion of our donor pool from a 5-day G-CSF protocol at 10 micrograms per kilogram per day to a shorter 4-day protocol at the same dose, following updated guidance from the Japan Society for Hematopoietic Cell Transplantation. That transition happened to coincide with the summer months of 2021 and 2022, which are also the months with the deepest dips in those two years. When we isolate just donors on the 5-day protocol across all five years, the summer dip is still present but slightly shallower in 2021 and 2022 compared to the pooled curve. The 4-day protocol cohort shows a more pronounced summer effect. This does not eliminate the seasonal signal — it appears in every year including 2019 and 2020 before the protocol change — but it adds a real confounding layer that we cannot fully disentangle retrospectively. It is a lesson in the cost of protocol changes that are not staggered across seasons deliberately.

Circadian timing and collection hour distribution across seasons

One variable we had not considered until a colleague flagged it during an internal review session: collection hour. Our facility runs collections between 8 a.m. and 4 p.m., and in summer the early slots fill faster because donors prefer to arrive before the peak heat of the day. Circadian variation in hematopoietic progenitor release is well documented in murine models, with CD34+ counts in peripheral blood peaking in the late afternoon. Whether that rhythm translates meaningfully to human peripheral blood after G-CSF mobilization, which flattens many normal oscillations, is not settled. When we stratify our data by collection hour in two-hour blocks, the early morning band (8 to 10 a.m.) does show marginally lower counts than the midday and afternoon bands across all seasons. In summer, 54 percent of collections fall in that early band versus 38 percent in winter. This is a real, quantifiable scheduling shift, and it likely contributes something to the observed dip, even if the magnitude is modest.

What this means for scheduling and clinical planning

The practical takeaway is not that summer collections should be avoided — the volumes are still clinically usable for the vast majority of recipients — but that labs and transplant coordinators working in Tokyo's climate should treat August intake counts as a baseline-adjusted figure. If a donor's pre-mobilization workup suggests they will sit near the lower threshold for adequate collection (typically 5 times 10 to the 6th CD34+ cells per kilogram of recipient weight), scheduling that collection in July or August without adjusting expectations is a setup for a borderline outcome. Our own practice since mid-2023 has been to flag any donor whose predicted yield falls within 20 percent of the minimum threshold and offer an October or November slot if the transplant timeline permits. In cases where summer collection is unavoidable, we have begun piloting a plerixafor add-on on day 4 for at-risk donors, which is consistent with current Japanese clinical guidelines on suboptimal mobilization.

What we still cannot explain, and where we are looking next

Roughly half the amplitude of the summer dip remains unaccounted for after controlling for hemodilution, collection hour, and protocol differences. Two hypotheses are on our list for the next phase of analysis. First, ultraviolet radiation exposure and its downstream effects on immune cell trafficking: Japan's summer UV index is among the highest in East Asia, and there is emerging evidence that UVB exposure modulates CXCR4-CXCL12 axis signaling, which governs progenitor cell retention in the bone marrow niche. Second, sleep disruption. Tokyo summer nights are hot and humid even with air conditioning, and poor sleep quality has documented effects on bone marrow output in human studies. We are designing a supplementary donor questionnaire for 2024 collections that will capture self-reported sleep quality and average outdoor exposure time in the week before mobilization. It is a small addition to the intake process, but if the correlation holds, it would help individualize mobilization risk scoring in a way that calendar month alone cannot.

Five years of data is enough to say with confidence that the summer dip in CD34+ counts at this laboratory is real, reproducible, and clinically worth accounting for. It is not yet enough to say exactly why it happens. That is the honest position, and it is the one that will shape the next phase of this analysis through 2025.