We have evaluated the specific effect of conditioning for HCT on the gonads. Any patient having other factors that might have interfered with their gonadal function was excluded, except the 27 given HCT in the second or third remission who were previously given chemotherapy without irradiation.
Their plasma FSH and LH concentrations indicated that 26% of our boys had normal testicular function. These figures are higher than those reported by Sanders et al  (13%) and Van Casteren et al  who reported that the 10 boys given TBI 7.5 or 12 Gy before HCT had almost undetectable (< 26 pg/mL, n = 8) or 26-62 pg/mL (n = 2) inhibin B concentrations; the severity of gonadal dysfunction in this study may be partly due to the additional testicular radiotherapy given to 5/10 boys. The differences might also be explained by differences in the age at HCT and/or the interval between HCT and evaluation. However, these two factors did not influence testicular function in our study. The wide ranges of the age at HCT and the interval between HCT and evaluation, and the limited study population might partly explain the absence of any significant impact of these factors in our study.
Melphalan seemed to be more gonadotoxic, giving rise to increased FSH (in all 16) and decreased inhibin B (in 14 of them) concentrations. This effect of melphalan was also seen by Crofton et al , who showed that inhibin B was normal before, during and after treatment in 16 boys given chemotherapy for solid tumors or acute lymphoblastic leukemia, except in one who had undetectable inhibin B after cyclophosphamide plus cisplatin plus melphalan. However conclusions cannot easily be drawn as the effect of the drug cannot be separated from that of irradiation.
Our study confirms the excellent correlation and concordance between the plasma concentrations of FSH and inhibin B that was reported by Cicognani et al . They showed that all boys who have increased FSH (> 6.1 IU/L) or LH (> 5.8 IU/L) concentrations also had inhibin B concentrations below 112 pg/mL (-2 standard deviations). Unlike Lähteenmäki et al , we found no correlation between testicular volume and inhibin B (P = 0.06). They reported that patients with small testes (< 10 mL) after treatment for childhood malignancy had inhibin B concentrations below 42 pg/mL, and 6 out of 7 had FSH concentrations above 9 IU/L; patients with a testicular volume above 13 mL had inhibin B concentrations above 100 pg/mL. This difference might be explained by variations in the age at which the boys in our study were given HCT and evaluated.
We find that the plasma AMH concentration does not seem to be a good marker, as it normally decreases when testosterone increases at puberty.
We did not carry out a sperm analysis because our boys were too young. However, the reported relationship between inhibin B and the sperm count in normal men and in boys surviving a childhood malignancy suggests that 10/38 (based on FSH or inhibin B) or 7/38 (based on both) of the boys we studied will be normospermic. Thomson et al  showed that inhibin B was barely detectable in azoospermic patients whose germ cells had been destroyed, despite preservation of their Sertoli cells, as confirmed by testicular biopsy. Van Beek et al  showed that inhibin B is a better marker of spermatogenesis than is the FSH concentration in men given chemotherapy for Hodgkin's lymphoma as children. Multivariate analysis with inhibin B and FSH concentrations as determinants of sperm concentration showed that inhibin B was the only significant determinant; all the men with a plasma inhibin B concentration above 75 pg/mL were normospermic. Lähteenmäki et al  showed that the inhibin B and FSH concentrations and the testicular volume explain 44% of the variance in the sperm count. Van Casteren et al  found that their 5 patients who were normospermic had inhibin B concentrations of 125-234 pg/mL, while the corresponding values in the 9 azoospermic patients was 0-79 pg/mL, with a correlation between sperm count and inhibin B concentrations.
Their plasma FSH and LH concentrations indicate that 21% of our girls had normal ovarian function. These figures are similar to those reported by others [2, 21, 22]. However, the girls in our study with spontaneous puberty were no younger at HCT than were those with ovarian failure. Similarly, the FSH, inhibin B and AMH concentrations were not influenced by age or pubertal status at treatment. Fong et al  studied women treated for cancer during childhood and found that those with undetectable AMH were significantly older at diagnosis, only 2/17 had regular menstrual cycles, and more of them had been given body or abdomen radiotherapy than those with greater AMH concentrations. The preeminence of the role of the irradiation and of busulfan might partly explain why the age at HCT has no significant impact on ovarian function in our study.
Busulfan seemed to be more gonadotoxic. Of the 8 girls given busulfan, combined with TBI in two, 7 had increased plasma FSH/LH concentrations and 7/8 had low inhibin B. This effect of busulfan was also seen by Teinturier et al , who showed that the 10 girls given busulfan (600 mg/m2) without TBI before autologous HCT all developed severe, persistent ovarian failure. They identified 26 of 29 girls in published reports given busulfan during prepuberty or puberty as having signs of ovarian failure.
Larsen et al  used multiple linear regression analysis of 100 girls who had survived childhood malignancy to predict the total number of antral follicles per ovary. Ovarian irradiation, alkylating chemotherapy, older age at diagnosis and a long period off treatment all reduced the number of follicles. Among the 10 girls conditioned for HCT by TBI, the 3 with spontaneous cycles had smaller ovaries, fewer total follicles and higher FSH than the others.
We have therefore confirmed the excellent correlation and concordance between the plasma concentrations of FSH and inhibin B, and agreement with the clinical evolution, while those of AMH were very low. The 7 girls with normal puberty and FSH and LH all had AMH concentrations similar to those with ovarian failure. Three [23, 26, 27] studies have evaluated the capacity of plasma AMH concentrations to assess the ovarian damage in adults surviving childhood malignancy. Bath et al  compared 10 girls treated for cancer during childhood, all with regular menstrual cycles, with 11 controls. The treated girls had smaller ovaries than the controls and significantly higher FSH and lower AMH concentrations, while the other hormone concentrations were unchanged. Thus, the AMH concentration indicates a depletion of ovarian reserve, despite regular menstrual cycles. Van Beek et al  showed that girls treated with MOPP (mechlorethamine, oncovin, prednisone, procarbazine) for Hodgkin's lymphoma during childhood developed into women with elevated FSH concentrations and decreased AMH concentrations. Three of the women with normal FSH had decreased AMH.