GENETICS OF SOMATIC CELLS 435 



When considering the question of isoantigenic variation in populations of neo- 

 plastic cells, mention has to be made of some findings on human patients with leukemia, 

 recently summarized and discussed by Salmon. 1146 Several patients have been des- 

 cribed, all with acute leukemias, in whom antigens of the A group have been modified 

 concurrently with the disease, leading to a mixture of two populations of erythrocytes, 

 consisting either of A 1 and 0, or of cells with normal and with weak A, respectively. 

 No other blood-group anomalies were present. In one case, the patient had been 

 grouped two years previously to the onset of leukemia as a normal A and only after 

 the disease developed did a mixture of normal and weak A appear. In another case 445 

 the abnormal, weak-J-group erythrocytes diminished in frequency, parallel to a remis- 

 sion of the leukemia. According to Salmon, 1146 chimerism could be excluded on genetic 

 and statistical grounds. All cases hitherto observed concerned agglutinability by 

 anti-/l antibodies. Salmon attributed them to somatic mutations, occurring in stem 

 cells that give rise to leukemic myeloblasts on the one hand and weak A mutants in 

 the red-cell series on the other hand. However, it may be questioned whether a change 

 in agglutinability really reflects a change in antigenicity. The work of Moller 887 on 

 the development of certain isoantigens in newborn mice shows that the sensitivity of 

 lymphoid cells to the cytotoxic action of isoantisera is subject to developmental changes, 

 immature cells being less or not at all reactive. This does not necessarily mean a 

 lack of antigens, since these can be demonstrated in absorption tests; the difference is 

 one of reactivity, connected with maturity, rather than one of antigenic specificity. 

 In this case, it is of interest to note that weak-yl erythrocytes in the case described by 

 Gold et a/. 445 continued to absorb antwl sera. Therefore it may be questioned 

 whether the observed changes are really mutational rather than developmental. In 

 the latter case, they would represent only another aspect of the immaturity of leukemic 

 cells (all cases so far described were acute leukemias) . This would also be in agreement 

 with the decrease of the abnormal group during remission. 



Before closing this discussion on tumor isoantigens, it seems appropriate to deal 

 briefly with the question of tumor-specific antigens in experimental tumors, that is, 

 antigens present only in the tumor cells but not in their host. For every such considera- 

 tion it is absolutely essential to distinguish critically between the isoantigens of the 

 histocompatibility system (always of importance in transplantation experiments in 

 which donor and recipient are not genetically identical) and true tumor-specific 

 antigens that are antigenic in the host of origin or in animals isogenic with it. For 

 critical evaluation, some more rigorous test of isogenicity is required than the mere 

 statement that an inbred strain was used. Skin grafting is probably the most reliable 

 test presently available. In experiments carried out with the intention of studying the 

 possible existence of host resistance against a tumor, it must be proved critically that the 

 donor animal in which the tumor has originated and the recipients of the graft do not 

 differ from each other with regard to their genetically determined isoantigenic composi- 

 tion and possible positive results are not due to more or less subliminal homograft 

 reactions. The experiments of Prehn and Main 1023 in particular have gone a long way 



