To the Editor:
In a recent issue of Blood, Haurie et al1 provided a detailed and historical review of cyclical neutropenia and other periodic phenomena in hematopoiesis. There are several points related to the review that need emphasis.
The feedback hypothesis as originally proposed postulated that hematopoiesis itself is an actual or potentially oscillatory system due to the presence of multiple and coupled negative feedback loops, each of which embodies a time delay and has nonlinear operating characteristics.2 This is essentially the conclusion drawn by Haurie et al. At the time it was presented, this hypothesis was a paradigm shift away from the concept that periodic phenomena in hematopoiesis were due to the action of external rhythms, such as hormonal rhythms. The feedback concept now appears to be taken as a given, because no other possibilities were even discussed in the article. However, while the feedback hypothesis can explain most oscillatory phenomena, it may not explain all. For example, the size of the peripheral lymphocyte pool and the length of lymphocyte life span are such that the oscillation that is occasionally seen in lymphocyte number is difficult to explain by the action of a feedback loop involving lymphocytes or by fluctuating input of lymphocytes from a cycling stem cell pool.
Stem cells may be involved in oscillatory phenomena in two ways, which are not mutually exclusive. Firstly, a stem cell defect may result in hematopoiesis moving into a region of actual or enhanced oscillation. Secondly, the stem cell compartment itself may be oscillatory. Both of these possibilities were postulated at the time that the feedback hypothesis was proposed and, contrary to a statement in the review, it was suggested that most cases of cyclical neutropenia result from a stem cell disorder.2 In my opinion, the distinction that is drawn between whether oscillation arises primarily as the result of a stem cell feedback loop or a peripheral feedback loop is artificial and hinders understanding. The system should be viewed as a whole, as a web of coupled feedback loops that interact with each other. Although a stem cell lesion is the most common disturbance in disease, oscillatory behavior may arise from disturbance at one of a number of points, as is shown by the situation in cyclical thrombocytopenia.
The control-systems aspects of hematopoiesis are very attractive to modellers. However, once the control-systems features of hematopoiesis are accepted, the ability to construct a model that shows oscillatory behavior, even if the model incorporates the latest advances in hematopoietic cell biology, really adds little new knowledge. Rather, the challenge to modellers would seem to be to provide detailed predictions for the input-output characteristics of the different parts of the various control systems so that these predictions can be tested by experimental hematologists and a truly quantitative description of hematopoiesis can emerge.
The oscillatory phenomena in chronic myeloid leukemia are occasionally of practical importance. If cycling is present but is not recognized, an increase of cytotoxic drug when the leukocyte count rises or a decrease when it falls may exacerbate the situation and be confusing. The appropriate course of action is to maintain a constant dose of cytotoxic drug and to carefully observe the situation.
Prof Morley is correct in pointing out that the “feedback hypothesis” for the origin of periodic hematological diseases (as clearly enunciated in his paper with Drs King-Smith and Stohlman) marked a major shift in our views of these fascinating disorders, and we were remiss in not pointing this out clearly and explicitly.
We agree with Prof Morley that the hematopoietic “system should be viewed as a whole, as a web of coupled feedback loops that interact with each other,” and that “the challenge to modelers would seem to be to provide detailed predictions (so) a truly quantitative description of hematopoiesis can emerge.” This is precisely the point of view that has emerged from our own research. Indeed, in our recent study of cycling in grey collies1-1 adopting this point of view was necessary to explain (from a modeling perspective) the data that we presented.