Effect of Chemotherapy on Skeletal Health in Male Survivors from Testicular Cancer and Lymphoma Free

Over recent decades, there has been a substantial improvement in cure rates and survival times in certain cancers. Over the same period, the use of adjuvant chemotherapy and hormone therapy has expanded greatly. Thus, there are increasing numbers of long-term survivors who have received these treatments. Whereas initial attention has naturally been focused on treatment of the malignancy, recent studies have highlighted the possible long-term detrimental effects of cancer treatments on the skeleton, with accelerated bone loss, which may lead to osteoporosis (1–4). The morbidity associated with osteoporosis, increased fracture rate, chronic pain, and increased hospitalization may have a major effect on quality of life.

Both clinical and preclinical studies on the growing skeleton have shown that single or multiagent chemotherapy in children may reduce bone growth and final height (5) and also bone mineral density (BMD) (6, 7). This reduced BMD during childhood may result in permanently reduced BMD in adulthood (8). Osteoporosis induced by chemotherapy treatment during adulthood has been studied more extensively in women than in men and the link between chemotherapy-induced premature ovarian failure and accelerated bone loss is well established (9). Hormonal therapies, such as tamoxifen in premenopausal women, may also contribute to bone loss (10), whereas newer agents, such as aromatase inhibitors, also enhance postmenopausal bone loss (11).

There is now a recognized need for more detailed evaluation of bone loss in men induced by cancer treatment during adulthood (2). In older men, treatment-induced bone loss is largely associated with hypogonadism resulting from androgen deprivation therapy for prostate cancer (12), with studies showing 0.6% to 9.6% and 2.3% to 4.6% annual reduction in BMD at hip and lumbar spine, respectively (4), whereas a very large population-based study confirmed the expected increased fracture rate in men receiving androgen deprivation therapy (13). However, in men with lymphoma and testicular cancer, the long-term effects of chemotherapy on bone loss are poorly defined. In lymphoma, small studies have suggested that, in a proportion of patients, bone loss may be associated with hypogonadism (12, 14), but large, more robust studies with modern chemotherapy regimens have not been carried out.

Accelerated bone loss may occur through direct effects of chemotherapy on bone, with methotrexate and ifosfamide suggested to have direct bone effects (5, 15). However, in clinical practice, most chemotherapy is given as a combination of several agents. It is therefore difficult to delineate individual effects. Chemotherapy is also often combined with steroid therapy, which will contribute to bone loss. The aims of the current study were therefore to assess the extent of possible bone loss due to previous chemotherapy in men and to determine if the rate of bone turnover in such patients is abnormal by measurement of bone metabolism markers.

Study design and patients. This was a single center, cross-sectional study, in which the BMD measurements of male patients, treated previously with chemotherapy with curative intent for lymphoma or testicular cancers, were compared with that of a male cancer control population that had not received chemotherapy. All consecutive patients attending lymphoma and testicular cancer follow-up clinics at Weston Park Hospital (Sheffield, United Kingdom), who were eligible, were invited to join the study. Written informed consent to participate in the study was required from all participants and the study had full approval of the South Sheffield Ethics Committee.

Inclusion criteria included males aged <70 years, with previously confirmed curable malignancy (lymphoma or testicular cancer), treated with chemotherapy, completed at least 12 months previously and in remission. The control patients were similar but had not received chemotherapy. All patients were aged ≥18 years on initiation of their cancer therapy. Exclusion criteria for both groups included patients with known metabolic bone disease.

Previous radiotherapy was not an exclusion criterion, but the site of treatment, dose, treatment time, and duration were recorded. It was assumed that localized radiotherapy would not have effects elsewhere on the skeleton and that the dose of radiation to the testes was minimal (<10 cGy). Previous studies in men treated with orchidectomy and radiotherapy for stage I seminoma of the testis that had measured BMD, including irradiated areas of bone, had shown no significant difference between irradiated and nonirradiated sites (16). No patients in our study received total body irradiation.

BMD measurements were made by dual-energy X-ray scanning (DEXA) using a Lunar EXPERT-XL imaging densitometer at the lumbar spine (L1-L4) and the proximal femur and for the whole body. The manufacturers of the EXPERT-XL have produced mean BMD values for 10-year intervals from a large reference population. This included 8,905 and 7,811 normal subjects for the spine and hip, respectively. These data correspond very closely to published reference data (17). The precision using the EXPERT-XL is 0.01 to 0.03 g/cm², depending on the site measured, with precision for the spine being greater than that for the proximal femur. DEXA measurements were also used to determine total body composition in terms of amount of trunkal fat and percent body fat.

The WHO criteria were used to define osteoporosis and osteopenia (18). This relates BMD in an individual to the mean BMD in normal healthy young adults, expressed in terms of a T-score, with osteoporosis defined as a T-score of less than −2.5 and osteopenia as a T-score between −1.0 and −2.5. A Z-score, relating BMD to a healthy age-matched population, was also calculated.

Patients completed a questionnaire on skeletal health and family history of hip fracture/osteoporosis. The results of the DEXA scans and issues raised by the questionnaire were discussed with patients and advice was given about lifestyle issues and, where appropriate, on bone-specific treatments.

Biochemical and hormone measurements. All samples from the chemotherapy group and nonchemotherapy group were analyzed in a randomized order by staff blind to group status and measurements were done using a single batch of reagents for each analyte. All patients attended for blood tests between 9:00 a.m. and 10:00 a.m. to minimize errors due to circadian rhythms and had the following assessments: full blood count, erythrocyte sedimentation rate, bone profile (including calcium), liver function tests, and routine electrolytes. This permitted screening for secondary causes of osteoporosis or other metabolic bone disease. Luteinizing hormone (LH), follicle-stimulating hormone (FSH), estradiol, and parathyroid hormone were measured by the appropriate Elecsys assay systems (Elecsys 2010 immunoanalyzer, Roche Diagnostics GmbH, Mannheim, Germany). Reference ranges were 1.7 to 12.2 IU/L for LH, 2.0 to 18.1 IU/L for FSH, 13.5 to 59.5 pg/mL for estradiol, and 12 to 70 pg/mL for parathyroid hormone. Testosterone was determined using the Vitros assay, reference range 7.1 to 24.1 nmol/L.

Bone marker measurements. Bone resorption was assessed by measurement of the urinary NH₂-terminal telopeptide fragment of type I collagen (NTX) in an early morning, second voided urine sample, collected on the day of outpatient attendance and stored at −20°C for later analysis. These measurements were made using a chemiluminescence assay for the NTX using a Vitros ECI analyzer and urinary creatinine measured using a dry slide method, in both cases using reagents from Ortho-Clinical Diagnostics (Rochester, NY). The NTX reference range for men used in our center was 16 to 107 nmol/mmol creatinine [coefficient variation, 6.8-14.9%]. Bone-specific alkaline phosphatase (BAP) was measured by immunoassay using the Access Ostase assay (Beckman Access, Beckman Coulter, Inc, Fullerton, CA). The reference range was 6.5 to 25μg/L, and coefficient variation was 4.0%.

Statistical analyses. The study by Holmes et al. (19) in lymphoma patients was used to determine the numbers of patients required. This study found deviations of BMD in terms of SDs below the mean of age-matched males (Z-scores) of −0.8 at the radius, −0.6 at the lumbar spine, and −0.4 at the femoral neck. Because the chemotherapy used was probably more aggressive on average than the chemotherapy in our study group, the smallest of these changes was taken as a basis to calculate sample size. Using the approach of Rigby et al. (20), for a difference of −0.4 SD units, it was calculated that at least 100 cases and 100 controls were needed (80% power, 5% significance, and two tailed).

Analyses were done with Stata version 7 (21) and Excel 2000 (22) and all tests were two sided. Regression analyses were used to test association between BMD and variables and between chemotherapy and variables. The relationship of variables to BMD (joint associations) was assessed by multiple regression with BMD as the response variable. To test whether variables were related to chemotherapy (a binary variable), a multiple logistic regression was done with chemotherapy as the response variable. The results of the analysis were expressed as odds ratios (OR) relating type of therapy to each explanatory variable. As this was an exploratory analysis, no correction for multiple comparisons was made and no imputation was done for missing data. Significance was taken as P < 0.05 (two sided).

Patient demographics. Patients were evaluated between May 2001 and August 2003. Two hundred seventeen men on standard follow-up in our center were recruited; 115 had received chemotherapy and 102 were cancer controls. The distribution of these patients by cancer type and prior treatments is shown in Table 1.

For the chemotherapy and control groups, the median ages were well matched at 40.4 and 42.3 years, respectively. The distribution of patient age according to decade is shown in Fig. 1 and is reasonably matched between the two groups. This is important because of the known variation of BMD with age in the normal population, although small differences can be adjusted for in the analysis. The median times since the completion of curative treatment were similar at 4.1 and 4.2 years, respectively.

The chemotherapy regimens received by the patients in the chemotherapy group are shown in Table 2, with BEP being the most common (see Table 2 for explanation of chemotherapy abbreviations). Ninety-eight patients received only first-line chemotherapy, whereas 17 patients went on to receive additional chemotherapy regimens. Nine of these latter patients (two testicular teratoma and seven lymphoma) received high-dose chemotherapy, with peripheral blood stem cell harvest. The other eight patients received only second-line chemotherapy. Only four patients in the chemotherapy group and two patients in the nonchemotherapy group received testosterone replacement therapy. The great majority of testis cancer patients had received orchiectomy, 58 of 64 (91%) and 99 of 101 (98%) of chemotherapy and nonchemotherapy groups, respectively.

From the lifestyle questionnaire (Table 1), it was clear that these patients represented an active and generally healthy population with little comorbid disease.

BMD as measured by DEXA scan. All patients in the study were categorized as normal, osteopenic, or osteoporotic using the T-score and Z-score according to the WHO criteria if these were met either at the femur or at the spine. Using T-scores, 23 (20%) patients in the chemotherapy group were found to be osteopenic compared with 17 (16.7%) patients in the cancer control group, with 2 (1.7%) patients osteoporotic in the chemotherapy group and no patient was osteoporotic in the control group. Using Z-scores, 17 (14.8%) patients were osteopenic in the chemotherapy group compared with 19 (18.6%) patients in the control group, whereas 2 (1.7%) patients were osteoporotic in the chemotherapy group, including 1 patient not osteoporotic on T-score (this can occur because of the different definitions of T-score and Z-score). One (1.0%) patient was found to be osteoporotic in the control group by Z-score. The ages of the osteoporotic patients in the chemotherapy group (on both T-score and Z-score) were 38, 46, and 63 years and the age of the patient in the control group was 32 years. The proportions of patients with osteopenia/osteoporosis were therefore broadly similar between the two groups.

The mean BMD values for all patients in the chemotherapy and control groups are shown in Table 3. Simultaneous adjustment by multiple regression for age at DEXA scan, time since last treatment, and body mass showed that there was no significant difference between the chemotherapy group and the control group for BMD at the femur (P = 0.347) or at the spine (P = 0.435). The mean BMD values for all patients in the chemotherapy and control groups according to decade of life are also shown in Table 3. When the relationship between age and BMD was tested, this was found to be significant for femur (P = 0.006) but not for spine (P = 0.196). For the testicular cancer subgroup (the great majority of whom received BEP; see Table 2), similar results were obtained (Table 3), with the BMD values for chemotherapy and control groups being almost identical at the spine (P = 0.680) and very similar at the femur (P = 0.622).

Further statistical tests were carried out for the whole patient population (adjusted for the same variables as above), between the chemotherapy and control groups to determine if there was a threshold time at which a difference might appear, on the basis that bone loss due to chemotherapy is a slow process. However, these tests found no evidence that BMD was affected by an interaction between chemotherapy and time since chemotherapy (P = 0.994).

For comparative purposes, the variation in BMD with age for both spine and femur are shown in Fig. 2, along with the BMD variation with age for the reference population provided by the manufacturer of the DEXA scanner (see Materials and Methods). This direct comparison also shows that there is no apparent difference between the chemotherapy and control groups and that neither of these shows lower BMD values than the reference population.

For the control group, a subanalysis showed that there were no significant differences in lumbar spine BMD between those subjects who had radiotherapy and those who received surgery alone (95% CI, 0.133-0.052; P = 0.389) after adjusting for body mass, age at DEXA scan, and time since last treatment. Thus, although the radiation field may have included an area of the lumbar spine, this has no significant effect on BMD, consistent with the conclusions of Stutz et al. (16).

For both spine and hip BMD values, in both chemotherapy and control groups, multiple linear regression showed that there was no significant association between BMD and estradiol, LH, FSH, or testosterone, but this should be seen against the small proportion of patients in this study found to have overt gonadal dysfunction.

There was no evidence that the low bone density found in a few patients was associated with receiving high-dose chemotherapy.

Bone turnover markers. The variation in NTX and BAP with decade is shown in Fig. 3 in box and whisker plot format. In all age groups, the median marker values were well below the top of the reference range for adult males (taken as 80 nmol/mmol creatinine for NTX and 25 IU/L for BAP). In the chemotherapy group, six (5.2%) patients had NTX values above the reference range compared with two (2.0%) patients in the control group. There was no association between NTX and the type or extent of chemotherapy. Logistic regression with NTX, BAP, and age at DEXA scan as covariates showed that there were no significant differences between the chemotherapy and control groups for NTX (OR, 1.00005; 95% CI, 0.9994-1.0007; P = 0.892). However, for BAP, the regression suggested a small but significant difference (OR, 1.009; 95% CI, 1.004-1.015; P = 0.001).

Multiple regression analyses showed that there were no significant associations between BMD measured either at the femur or spine and either NTX or BAP levels (again adjusting for age at DEXA scan).

Body composition. Total body mass was highly significantly related to BMD at both spine and femur (P < 0.001 in both cases). This is a well-known relationship but acts as a good check on the methodology. Logistic regression analyses of trunkal fat, trunkal lean, peripheral fat, and peripheral lean showed no significant differences for chemotherapy patients versus controls with all four of these body mass variables as covariates.

Hormone levels. Simultaneous analyses of sex hormone levels by multiple logistic regression showed that there were no significant differences between the chemotherapy and control groups for estradiol, LH, and testosterone (Ps and 95% CIs are shown in Table 4) but that the FSH values were significantly higher in the chemotherapy group compared with the control group (P = 0.011). Parathyroid hormone levels in the chemotherapy group [37.9 ± 17.2 (SD) pg/mL] and control group [36.4 ± 15.2 (SD) pg/mL] were also similar. In the chemotherapy group, 15 (13%) patients had testosterone levels below 7.1 (the lower end of the reference range). Nine of these patients were in the testicular cancer group and six were in the lymphoma group. In the control group, 10 (10%) patients had low testosterone and all were in the testicular cancer group.

This study is believed to be the largest evaluation of bone health in men who have received curative treatment for cancer. The key observations are that platinum-based chemotherapy for testicular cancer and anthracycline/alkylating agent-based combination chemotherapy for lymphoma in men have no significant effects on BMD or bone metabolism. With >100 patients in both chemotherapy-treated and cancer control groups, we can be confident that there was <0.4 SDs or 3% in BMD between the groups. However, we were only able to evaluate nine patients treated with high-dose chemotherapy and thus cannot provide reliable reassurance for this subset of patients, where one might expect the bone marrow ablation associated with high-dose chemotherapy to have an effect on bone cell precursors and function (23).

Although, Holmes et al. (19) showed small reductions in BMD of the spine, hip, and forearm in 29 men treated with combination chemotherapy for Hodgkin's lymphoma compared with age-matched male controls, endocrine causes were not clearly excluded. Howell et al. (24) showed significant reduction in BMD in a group of 36 men who had received procarbazine-containing chemotherapy or high-dose chemotherapy for hematologic malignancy. However, these men also had evidence of mild Leydig cell impairment that may have contributed to the findings.

Because any effects on BMD could be associated with the endocrine effects of treatments, it was important to assess sex hormone levels in this study. For estradiol, LH, and testosterone, there were no differences between the chemotherapy and control groups. However, FSH was found to be significantly higher in the chemotherapy-treated patients, consistent with the known effects of chemotherapy on the germinal epithelium, with the damage reflected by an increase in FSH levels and/or reduced sperm counts (25, 26). There is some interaction between sperm production by Sertoli cells under the control of FSH and testosterone production by Leydig cells under the control of LH, but this increased FSH alone is not likely to have a major effect on testosterone production. Our data on testosterone levels would tend to support this assumption.

The finding that there was no difference between chemotherapy and nonchemotherapy groups for bone resorption and that this seemed to be normal in both groups gives further confidence that there was no chemotherapy-associated accelerated bone loss. The reason for the apparent association in the statistical regression between chemotherapy and bone formation (with bone formation being greater in the chemotherapy group) is not clear.

Was the use of a cancer control group appropriate? We believe so, as there are data to suggest that cancer itself can affect bone health (27) and the aim of our study was to specifically evaluate the effects of chemotherapy on normal bone. However, we also compared the BMD results in the cancer controls with the age-matched healthy male reference population provided with the Lunar EXPERT-XL densitometer. This analysis also showed no evidence of a reduction in BMD in the control group in this study compared with the data from age-matched healthy subjects, providing reassurance that the lack of any difference in BMD between the chemotherapy group and the control group was not due to bone loss in the control group itself. Interestingly, there is a suggestion of a marginal increase in the BMD of patients in this study compared with the healthy population (reported in 1995), and, although we could not show that this difference was significant, it is consistent with recent findings that mean BMD of healthy populations has increased in recent years with improved nutrition (28).

There are several reasons that might explain our findings, when there have been suggestions of bone loss in the previous small studies of lymphoma patients. Part of the explanation may lie in the less aggressive chemotherapy used, on average, in our patients. Apparently, testicular cancer patients in particular are less at risk of chemotherapy-induced accelerated bone loss than other groups. This could be linked with their low mean age and their lifestyle following curative treatment. Many of the patients in this study are in full-time employment and activity and exercise levels remain high.

This is consistent with findings in other studies (29). The possibility of recovery from a chemotherapy-induced bone loss should also be considered (i.e., some patients may experience an accelerated reduction in BMD soon after chemotherapy but, in this younger age group, may subsequently be able to restore their BMD at least partially). A prospective longitudinal study would be worthwhile to characterize any short-term changes and subsequent recovery after treatment is complete.

These results have practical implications for the management of patients with lymphoma and testicular cancer and the possible role of osteoporosis screening. We suggest that it is probably unnecessary to routinely screen all testicular cancer and lymphoma patients who have received chemotherapy and that such screening should be reserved for those who have particular risk factors (e.g., family history of osteoporosis or history of low trauma fracture).

In conclusion, the finding of no detectable effects on BMD and bone turnover of chemotherapy in this group of patients is reassuring for the increasing numbers of male cancer survivors. These patients are typically relatively young men in whom accelerated bone loss leading to osteoporosis would have been a burden for many years. Although it is always wise to treat the results of a single study with caution, the relatively large numbers of patients in this study and the breadth of the associated measurements lend confidence to our findings.