IN CHILDHOOD

The degree of long term lymphopenia induced by cranial irradiation is shown to be dependent upon the number of fractions into which the standard cranial dose of 2400 rads is divided. The mean lymphocyte count of patients three months after receiving this dose in five fractions was 1.84 X 108/1; in twelve fractions it was 1.12 X 109/1 and in twenty fractions it was 0.64 X 109/l. An explanation for this finding is offered which is based upon marked radio-sensitivity of certain lymphocytes and the pattern of recirculation and redistribution of these cells

higher rate of haematological relapse. ' Cranial irradiation with intrathecal methotrexate does result in less long-term lymphopenia than craniospinal irradiation; both types of treatment leave the patients with a highly significant reduction in lymphocyte counts for at least two years. lo In this paper we show that the level of longterm lymphopenia induced by a given total dose of cranial irradiation is dependent upon the number of fractions into which the irradiation is divided. The cranial fields irradiated in leukaemia contain very little lymphoid tissue. The occipital, mastoid and superficial nodes and some adenoidal tissues are directly in the field, but the tonsils are not included. Consequently the main bulk of lymphocytes entering the fields during irradiation are those circulating in the blood. It is, therefore, unlikely that more than 10% of lymphocytes would be exposed in any one fraction; even if all these were destroyed 90% of the body pool would remain. However, in the period between fractions there would be almost complete reequilibration of the remaining 90% so that at the next session 10% of this 90% would be irradiated. If lymphocytes which are lost for a long period following irradition are very radiosensitive, then the number of lymphocytes killed in any one irradiation session would be largely independent of the dose given in that session, Le., if 100 rads kill 90% of susceptible lymphocytes only 10% more can be lost if the dose is 108 0008-543X-78-0100-0108-0060 @ American Cancer Society Oral MTX = methotrexate by mouth or intramuscular. In some patients in UKALL 111 the 6-mercaptopurine dose was reduced before and during cranial irradiation. This is unlikely to have affected the lymphocyte counts at 24 weeks. .4sparaginase was included in both trials and was a randomised variable in UKALL 111. It is not included in this figure for clarity.
Other groups in UKALL I1 received craniospinal irradiation. Another group in UKALL 111 received more intensive chemotherapy. These other groups are not considered here. M. P .

IT M T X O R A L M T X C R A N I A L I R R A D I A T l O N
increased many times. Under these circumstances the degree of lymphopenia (expressed as log lymphocyte counts) induced by a given total cranial dose will be linearly related to the number of fractions into which the total dose is divided. The British Medical Research Council UKALL I1 and UKALL 111 trials of treatment of ALL specified that children should receive cranial irradiation to a total dose of 2400/2500 rads after remission induction, but did not indicate the number of fractions in which this total dose should be given. Consequently different numbers of fractions, from 5 to 20, were given at different cooperating centres. This has made possible the analysis of the effect of this variation on the long term lymphopenia induced.

PATIENTS AND METHODS
All patients receiving cranial irradiation in the MRC's UKALL I1 and 111 trials were included in the analysis. Details of treatment are given in Fig. 1. Cranial irradiation was given by opposing lateral fields in the 7th, 8th and 9th weeks after starting treatment. The apparatus varied from center to center and included supervoltage, 250 KV and cobalt sources. With the exception of patients receiving irradiation in five fractions, almost all patients were irradiated on a five-days-a-week basis, so that ten fractions were completed in 12-14 days, twelve fractions in 16 days, fifteen fractions in 19-21 days and twenty fractions in 26-28 days. The patients receiving irradiation in five fractions had this in 16 days. The total protocol specified that total and differential leucocyte counts should be carried out where possible on the first day of each treatment week. All counts which were verified as being at the start of treatrnent week 24 in UKALL 111 and treatment week 22 in UKALL I1 were included in the analysis. The justification for pooling data from the two trials is given in Results. Log lymphocyte counts are used, as the distribution of these approximates well to the normal; median counts are very close to log means. We have previously shown that where large numbers of patients are studied the inaccuracies of the conventional lymphocyte or neutrophil count are lost and considerable precision is obtained. lo,ll*ls

RESULTS
The most frequently chosen number of fractions into which the 2400/2500 rads cranial irradiation was divided was 15. Analysis of mean log lymphocyte counts for the two trials, using only patients who had had 15 fractions, showed that the point at which the two schedules were most similar was at the beginning of week 22 in UKALL I1 and week 24 in UKALL 111. This point represented the maximum lymphopenia in the first treatment cycle for both trials. The lymphocyte count data for the two trials at this stage are given in Table 1.

CANCER January 1978
Vol. 41 Week 22 Week 24 It will be seen that the standard deviation is quite large for these groups but the median characteristically falls close to the log mean count.
By taking all the data at week 22 in UKALL I1 and week 24 in UKALL 111, 119 observations were available: five at 5 fractions, eight at 10 fractions, three at 11 fractions, thirty-one at 12 fractions, four at 13 fractions, fifteen at 14 fractions, forty-one at 15 fractions, one at 16 fractions, one at 17 fractions and ten at 20 fractions. The log lymphocyte count was then plotted against the number of fractions of irradiation given and the result is shown in Fig. 2. (The mean values for week 11, 13, 16 and 17 are not drawn as there were so few patients in these groups, but these data were used in calculating the correlation coefficient.) There is a linear relationship between these values with a correlation coefficient of 0.31 p < 0.001. As the number of fractions of irradiation the patients had received was determined by the center at which they were attending, there could have been a considerable bias in terms of age distribution and consequently expected normal blood lymphocyte count. In order to see if this was the case, blood lymphocyte counts from all the patients analyzed in Fig. 2 were collected during the period after they had achieved remission but before they had started CNS prophylaxis; the log mean values are shown in Fig. 2.
They show no intrinsic bias to low counts in the patients who were to receive many fractions of irradiation or to high counts in those that were to receive few fractions.

DISCUSSION
Ford and Hollingsworth carried out experimental irradiation, which in many ways parallels the observation we describe. '9' In their experiments they attached a &emitting 32Pimpregnated celluloid strip to either the spleen or appendix of laboratory animals. The pathlength of this isotope could only have destroyed lymphocytes passing within one or two mm of the strip and yet gross depletion of the recirculating pool of lymphocytes was seen after the strips had been in place some days. These experiments strongly suggested that most recirculating lymphocytes had passed through the irradiation field.
In the rat it has been estimated that only about 5% of recirculating T cells are in the blood at any one time while approximately 25% are in the spleen and 70% in the lymph nodes, lymph, gut-associated lymphoid organs and other tissues. 6 It is interesting that each fraction of irradiation in Fig. 2 carried a mean fall of between 5 and 6% in the peripheral blood lymphocyte count three months later.
If the data we have presented are taken at face value then one must conclude that the lympho- cytes which are being depleted as a result of irradiation and which failed to recover within three months are very radiosensitive. This follows from the finding that as large a proportion of lymphocytes are lost as the result of a single 120 rad fraction as of a 240 or 280 rads fraction. It has already been argued that most lymphocytes are irradiated in the blood vessels of the head. As not more than one-fifth of the blood volume is in the cranial field, each blood lymphocyte on average would receive no more than one-fifth of the total rad given to the cranium, assuming total mixing of the blood during irradiation. This would indicate that as little as 24 rads is sufficient to destroy the most radio-sensitive lymphocytes. We have previously shown in studies of patients receiving spinal irradiation that there is a plateau above which a considerable increase in irradiation does not significantly increase long term lymphopenia. l3 These data indicated that the mean size of the pool of lymphocytes which was lost for a protracted period following irradiation was represented by about 1.2 X lo9 lymphocytes/liter. O n the basis of this observation one would expect that the line drawn in Fig. 2 should flatten as it approaches 0.5 X lo9 lymphocytes/liter. There are no data on patients given more than 20 fractions so this postulation cannot be tested here.
The final point to consider is whether there is any advantage to giving a smaller number of fractions apart from lessening the long-term lymphopenia induced by CNS prophylaxis.
Clearly there is an economic advantage in terms of hospital, patient and ambulance time. The information which we have available in this retrospective trial fails to show any obvious difference of relapse rate in the CNS or elsewhere, of infections or of death in remission between the patients. But the two groups are too small to be certain that such differences would not be revealed in a larger study.
Nor are there as yet analyzable data on growth rate. Shalet et a1. I2 have shown that the dose of cranial irradiation required to cause pituitary damage resulting in loss of growth-hormone production is very close to that given in this study. They indicate that irradiation given in fewer fractions may result in a larger effective dose to the pituitary leading to failure of hormone production. This finding and the possible hazard of cerebral damage indicate that the adoption of CNS irradiation regimens with reduced fraction numbers should be made with caution. It would be wise to refrain from using fractions larger than 300 ratls until the longterm effects are more precisely known. However, as fewer fractions result in a greater effective dose it should be possible to reduce the total irradiation as fraction numbers are reduced without loss of antileukaemic effect, assuming leukaemic cells in the CNS are stationary. The MRC Working Party is currently comparing the effectiveness and complications of 2100 rad cranial irradiation given in 7 fractions with 2400 rad given in 12 fractions.