Rapid-onset hypogonadism secondary to crizotinib use in men with metastatic nonsmall cell lung cancer


  • Andrew J. Weickhardt MBBS, DMedSc,

    Corresponding author
    1. Division of Medical Oncology, Department of Medicine, University of Colorado Anschutz Medical Campus, Aurora, Colorado
    • Mail Stop 8117, 12801 East 17th Avenue, Aurora, CO 80045

    Search for more papers by this author
    • Fax: (303)724-3889

  • Micol S. Rothman MD,

    1. Division of Endocrinology, Department of Medicine, University of Colorado Anschutz Medical Campus, Aurora, Colorado
    Search for more papers by this author
  • Smita Salian-Mehta MD,

    1. Division of Endocrinology, Department of Medicine, University of Colorado Anschutz Medical Campus, Aurora, Colorado
    Search for more papers by this author
  • Katja Kiseljak-Vassiliades MD,

    1. Division of Endocrinology, Department of Medicine, University of Colorado Anschutz Medical Campus, Aurora, Colorado
    Search for more papers by this author
  • Ana B. Oton MD,

    1. Division of Medical Oncology, Department of Medicine, University of Colorado Anschutz Medical Campus, Aurora, Colorado
    Search for more papers by this author
  • Robert C. Doebele MD, PhD,

    1. Division of Medical Oncology, Department of Medicine, University of Colorado Anschutz Medical Campus, Aurora, Colorado
    Search for more papers by this author
  • Margaret E. Wierman MD,

    1. Division of Endocrinology, Department of Medicine, University of Colorado Anschutz Medical Campus, Aurora, Colorado
    2. Research Service, Veterans Affairs Medical Center, Denver, Colorado
    Search for more papers by this author
  • D. Ross Camidge MD, PhD

    1. Division of Medical Oncology, Department of Medicine, University of Colorado Anschutz Medical Campus, Aurora, Colorado
    Search for more papers by this author



The objective of this study was to document the differences in testosterone (T) levels between crizotinib-treated and noncrizotinib-treated patients with metastatic nonsmall cell lung cancer (NSCLC).


Testosterone levels were measured in 19 men with metastatic NSCLC who received crizotinib and in 19 men with metastatic NSCLC who did not receive crizotinib. Clinical characteristics of the patients were compared, and additional hormone assays were performed as appropriate. Two patients who began crizotinib and 4 patients who had dose interruptions or who stopped crizotinib therapy had serial hormone measurements, permitting the documentation of dynamic hormone changes on and off crizotinib treatment.


Total T levels were low (<241 ng/dL) in 19 of 19 (100%) crizotinib-treated men and in 6 of 19 men (32%) with metastatic NSCLC who did not receive crizotinib (mean T levels, 131 ng/dL and 311 ng/dL, respectively; P = .0002). Only 1 in 5 patients who had anaplastic lymphoma kinase (ALK) gene rearrangements and had not yet received crizotinib had low T. The initiation of crizotinib in 2 patients who had previously normal T levels was associated with a rapid decreases in T and in luteinizing hormone and follicle stimulating hormone levels within 14 to 21 days. Discontinuation of crizotinib led to increases back to normal T levels.


Crizotinib therapy caused rapid suppression of T levels in men. The current results indicated that the site of action must include a central (hypothalamic or pituitary) effect, but additional direct testicular effects could not be excluded. Further work is required to assess the correlation between low T levels and crizotinib side effects as well as the exact molecular mechanism and site of drug toxicity. Cancer 2012. © 2012 American Cancer Society.


Approximately 4% of patients with nonsmall cell lung cancer (NSCLC) have rearrangements in the anaplastic lymphoma kinase (ALK) gene, leading to an oncogene-addicted state from aberrant ALK activation.1-9 Crizotinib, an oral ALK and MET-directed tyrosine kinase inhibitor, has produced a high response rate and prolonged median progression-free survival in patients with ALK-positive NSCLC.10 Side effects of crizotinib include grade 1/2 nausea (54%), diarrhea (48%), visual disturbance (41%), constipation (24%), dizziness (15%), and fatigue (10%).10 Crizotinib (Xalkori; Pfizer, New York, NY) recently received US Food and Drug Administration approval for the treatment of patients with advanced, ALK-rearranged NSCLC, a group that potentially represents >10,000 patients per year in the United States alone.11

The potential for crizotinib-related hypogonadism came to our attention through an index case. A man aged 35 years with metastatic NSCLC and an ALK rearrangement who received crizotinib (250 mg twice daily in 28-day cycles) for 14 cycles reported worsening fatigue and sexual disinterest. His most recent computed tomography scan demonstrated an ongoing objective response. A complete blood profile, metabolic panel, thyroid-stimulating hormone level, and total testosterone (T) level were assessed. Although the other blood results were normal, the T level was significantly low (130 ng/dL; normal range, 241-850 ng/dL). The patient was referred to endocrinology, which highlighted recent data on hypogonadism in men with metastatic cancer patients being under diagnosed, under treated, and correlated with reduced quality of life (QoL).12 On the basis of this information, from early 2011, we adopted a policy of offering standard-of-care T testing to all men with metastatic cancer within the Thoracic Oncology Program to identify and refer patients who might benefit from testosterone therapy.

Although we identified a frequency of hypogonadism similar to that reported in the literature in our general NSCLC population, the crizotinib-treated population was particularly affected.13-17 The incidence of low T in men who were receiving crizotinib therapy and the observed rapidity of its onset and reversibility on withholding of the drug are reported here, and its consequences and possible basis are discussed.


The University of Colorado Thoracic Oncology Program adopted a policy of screening all patients with NSCLC who had tissue available for an expanding series of molecular abnormalities from mid-2008.18 Testing is conducted on-site within the Clinical Laboratory Improvement Amendments-certified Colorado Molecular Correlates (CMOCO) laboratory. An institutional review board-approved protocol permits clinical correlates to be made on all in-house patients in whom molecular analyses have been conducted within the CMOCO laboratory. After identification of the index case and recognition of the recently reported high incidence of low T levels in patients with advanced cancer, all men with metastatic NSCLC were offered T testing as part of routine clinical care. Because of our involvement in the early clinical trials of crizotinib, a large group of crizotinib-treated patients (the crizotinib-treated group [CTG]) was available for analysis. Crizotinib-treated patients received crizotinib at a dose of 250 mg twice daily in 21-day or 28-day cycles. Measurement of T in patients with metastatic NSCLC who had never received crizotinib (the noncrizotinib-treated group [NCTG]) served as a relevant comparator group. For initial comparison purposes, only patients with NSCLC who had ALK rearrangements and were tested for T before receiving crizotinib but who subsequently received crizotinib were included in the NCTG.

Testosterone levels were measured using a chemiluminescent immunoassay on the ADVIA Centaur XP platform (Siemens, Washington, DC) at the University of Colorado Hospital. Samples were obtained at clinic visits (8am to 4pm) rather than at any consistent time of day. Patients with proven or suspected hypogonadism (on the basis of symptoms) had additional hormone assessments to further explore the site of endocrine dysfunction.19, 20 Follicle-stimulating hormone (FSH), luteinizing hormone (LH), and prolactin (PRL) measurements were performed using chemiluminescent immunoassays on the ADVIA Centaur XP platform (Siemens). Several patients who had low T had additional levels measured using liquid chromatography-tandem mass spectrometry followed by radiolabeled T/equilibrium dialysis to calculate the percentage of free T. In the initial comparison, when T was measured more than once in the same patient, only the first assay result was used.

Patient age, current NSCLC-directed therapy, and previous therapy in terms of the total number of cycles of separate cytotoxic chemotherapy and the number of cycles of erlotinib or gefitinib therapy (1 cycle = 28 days) were recorded. Patients were classified as untreated at the time of blood sampling if they had not received chemotherapy or targeted therapy within 28 days before the T assessment. The presence or absence of brain metastases was documented as well as the site of any intracranial metastases. If no brain magnetic resonance imaging studies had been obtained in the 6 months before the measured T level, then this field was documented as unknown. Previous central nervous system-directed therapy with whole-brain radiation treatment (WBRT) or stereotactic radiation also was recorded. Current opiate and dexamethasone use (regular or as needed) at the time of the T assay and serum albumin levels also were documented. Information on fatigue, graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (NCI-CTCAE) (version 3.0) was available only for crizotinib-treated patients through a retrospective review of clinical trial documentation. Statistical significance in mean levels of T, FSH, LH, and PRL between the CTG and the NCTG was calculated using a 2-tailed, unpaired t test with Prism software (GraphPad Software, San Diego, Calif).


Patient Demographics

Demographic characteristics of the patients with metastatic NSCLC who were tested for random T levels are listed in Table 1. Nineteen patients who received crizotinib treatment for metastatic NSCLC were tested, including 17 patients with ALK gene rearrangements, 1 patient with a c-ros oncogene 1, receptor tyrosine kinase (ROS1) fusion gene, and 1 patient with MET gene amplification. Three ALK-positive patients within this group had T levels assessed both before and after the initiation of crizotinib therapy. Nineteen patients with metastatic NSCLC who did not receive crizotinib therapy were tested, including 5 patients with epidermal growth factor receptor (EGFR) mutations and 2 patients with ALK gene rearrangements (both were crizotinib naive and were receiving pemetrexed-based therapy at the time). The majority of patients in the NCTG were receiving chemotherapy or erlotinib (14 of 19 patients; 74%).

Table 1. Demographics of Patients With Metastatic Nonsmall Cell Lung Cancer Who Were Tested for Random Testosterone Levels
 No. of Patients (%)
VariableCrizotinib TreatedNoncrizotinib Treated
  1. Abbreviations: ALK, anaplastic lymphoma kinase; BMI, body mass index; Chemo cycles, sum of individual cycles of cytotoxic chemotherapy; EGFR, epidermal growth factor receptor; KRAS, v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog; MET, met proto-oncogene (hepatocyte growth factor receptor); py, pack-years; ROS1, c-ros oncogene 1, receptor tyrosine kinase; WBRT, whole-body radiotherapy.

Total no. of patients19 (100)19 (100)
Age: Median[range], y60 [35-81]63 [43-78]
 Adenocarcinoma19 (100)16 (84)
 Squamous cell0 (0)1 (5)
 Large cell0 (0)2 (11)
Molecular status  
 ALK rearranged17 (90)2 (11)
 ROS1 fusion gene1 (5)0 (0)
 MET amplification1 (5)0 (0)
 EGFR mutation0 (0)5 (26)
 KRAS mutation0 (0)3 (16)
 Undefined/other0 (0)9 (47)
Smoking status  
 Never/light smoker: <10 py15 (79)7 (37)
 Heavy smoker: >10 py4 (21)12 (63)
Prior therapy  
 Chemo cycles: Mean [range]19 [6-48]9 [2-38]
 Lines of therapy, mean no.1.81.5
Current therapy  
 Crizotinib19 (100)0 (0)
  Median duration, mo7.3
 Erlotinib0 (100)4 (21)
 Chemotherapy0 (100)10 (53)
 Untreated in last 1 mo0 (100)5 (26)
Brain metastases  
  Present5 (26)5 (26)
  Absent10 (53)10 (53)
  Unknown4 (21)4 (21)
Prior WBRT3 (16)3 (16)
Albumin: Mean, g/dL3.13.3
BMI: Mean, kg/m226.524.4
Current opiate use3 (20)4 (18)

Patients in the CTG and the NCTG had a similar median age (60 years and 63 years, respectively), similar rates of brain metastases (26% each) assessed by magnetic resonance imaging within the 6 months before the first T assessment, and a similar frequency of receiving previous WBRT (16% each) (Table 1). In the CTG, 8 of 19 patients (50%) had grade 1 or 2 fatigue, and there were with no documented occurrences of grade 3 or 4 fatigue. NCI-CTCAE–graded symptoms were not available for the NCTG. No patient in either group was receiving concurrent corticosteroids.

Hormone Measurements

Testosterone levels were significantly lower in the CTG compared with the NCTG (mean total T level, 131 ng/dL vs 311 ng/dL; P = .0002) (Fig. 1). In the CTG, 19 of 19 patients (100%) had low T (<241 ng/dL) compared with 6 of 19 patients (32%) in the NCTG. Of 5 patients with ALK-rearranged NSCLC who had T levels measured before receiving crizotinib (including 2 patients who remained on pemetrexed therapy at the time of the current analysis and have not yet received crizotinib therapy and 3 patients who subsequently went on to receive crizotinib treatment), only 1 patient (20%) had low T (mean, 355.8 ng/dL). Free T levels also were low (<9 ng/dL) in 9 of 10 patients tested in the CTG (mean, 6.8 ng/dL). No patients in the NCTG had free T levels measured. There was no correlation between the time of day blood was drawn for the hormone assay and the T level in patients (r = −0.04).

Figure 1.

A comparison of total testosterone between the crizotinib-treated and noncrizotinib-treated groups is illustrated. The red dashed line indicates the mean total testosterone level in the crizotinib-treated group; blue dashed line, the mean total testosterone level in the noncrizotinib-treated group.

Measurements of FSH, LH, and PRL were available for 13 of 19 patients in the CTG with low T and for 4 of 6 patients in the NCTG with low T (Table 2). Among the patients who had low T levels, there were no significant differences between the CTG and the NCTG in mean levels of FSH, LH, or PRL. Thyroid-stimulating hormone measurements were available for 13 of 19 patients with low T in the CTG and in 4 of 6 patients with low T in the NCTG, and were all within normal limits.

Table 2. Differences in Follicle-Stimulating Hormone, Luteinizing Hormone, and Prolactin Levels in Patients With Low Testosterone in the Crizotinib-Treated and Noncrizotinib-Treated Groups
VariableCrizotinib TreatedNoncrizotinib TreatedStatistical Significance
  • Abbreviations: FSH, follicle-stimulating hormone; LH, luteinizing hormone; NS, nonsignificant; PRL, prolactin.

  • a

    Thirteen of 19 patients who received crizotinib treatment and had low testosterone levels had FSH, LH, and PL measurements.

  • b

    Four of 6 patients who did not receive crizotinib treatment and had low testosterone levels had FSH, LH, and PL measurements.

No. of patients tested13a4b 
FSH, mIU/mL   
 Mean (range)17.4 [5-46]13.0 (5-32)NS
 Normal range1-181-18 
LH, mIU/mL   
 Mean (range)5.7 [1.4-10.2]6.0 (1.8-12)NS
 Normal range1.5-9.31.5-9.3 
 Mean (range), ng/mL13.0 (6-26)16.0 (5-34)NS
 Normal range, mIU/mL3-133-13 

Dynamic Changes Associated With Crizotinib Commencement

Two patients with ALK-rearranged NSCLC who previously had normal T levels had documented, rapid decreases in T within 14 to 21 days of initiating crizotinib therapy (Fig. 2a). Levels of LH, which were elevated at baseline, decreased in both patients over the same period (Fig. 2b). FSH levels, which also were elevated at baseline, decreased in 1 patient but remained stable in the other patient (Fig. 2c).

Figure 2.

Changes in total (a) testosterone, (b) lutenizing hormone (LH), and (c) follicle-stimulating hormone (FSH) levels at the start of crizotinib treatment (cycle 1, day 1 [C1D1]) are illustrated. Solid lines indicate Patient 1; dashed lines, Patient 2. LLN indicates lower limit of normal; ULN, upper limit of normal.

Four patients in the CTG either temporarily held crizotinib or permanently discontinued dosing because of toxicity (neutropenia), concurrent radiotherapy, or progression, allowing for additional dynamic assessment of T levels (Fig. 3). On each occasion, T levels increased within days of cessation of crizotinib and, conversely, decreased within days after reinstitution of crizotinib (Fig. 3). Simultaneous gonadotropin levels were not measured in these patients.

Figure 3.

Dynamic changes in testosterone associated with crizotinib dose delay or cessation are illustrated. Crizotinib was given orally at a dose of 250 mg twice daily, and each cycle (C) lasted 28 days (D). RTx indicates radiotherapy.

All patients with low T attended the Endocrinology Department to discuss the benefits and risks of testosterone therapy, and 10 of 19 patients subsequently commenced T therapy. When physiologic T in the form of topical gel was instituted, T levels increased to the normal range despite ongoing crizotinib therapy (data not shown). No formal documentation of the frequency or NCI-CTCAE–graded severity of specific hypogonadism-related symptoms was made before or after patients started T therapy.


On the basis of the identification of an index case of low T in a patient with NSCLC with an ALK gene rearrangement who received crizotinib treatment, and based on an increased recognition of the high frequency of hypogonadism in men with advanced cancer, we performed T measurements in patients with metastatic NSCLC to identify and refer patients who might benefit from physiologic T therapy.

Crizotinib treatment was associated with a 100% incidence of low T in 19 patients tested, compared with 32% of patients with metastatic NSCLC who did not receive crizotinib (Fig. 1). Patients in each group had similar use of opiates and dexamethasone and similar nutritional status as assessed by serum albumin and body mass index. In addition, there was no difference between the groups in the receipt of antiemetics (data not shown). Patients in the CTG were slightly younger, excluding an age-related impact on T levels as a major contributor to our observations.21 The CTG was more likely to have received more cycles of prior chemotherapy, which may be associated with lower baseline T because of direct gonadal toxicity from the chemotherapeutic drugs (Table 1).17, 22 Only 1 of 5 ALK-positive patients had low T levels when they were not receiving crizotinib, excluding the molecular subtype of cancer as the primary contributor to this observation. Although there is a substantial background incidence of hypogonadism in patients with metastatic NSCLC,12, 13 the rapid dynamic changes in T with decreases after commencement and increases after cessation of treatment with crizotinib in the CTG unequivocally demonstrate a direct causal relation between the crizotinib and effects on the hypothalamic-pituitary-testicular axis (Figs. 2, 3).

Detailed analysis of the 2 patients in whom T, LH, and FSH levels were tracked prospectively at the commencement of crizotinib offers some initial insights into the potential site(s) of drug toxicity (Fig. 2). Both LH levels and FSH levels were above the upper limit of normal before crizotinib was commenced, suggesting some prior gonadal damage (possibly from prior chemotherapy) and a requirement for increased gonadotropin drive to maintain normal T levels. After the initiation of crizotinib, although T levels rapidly fell, no compensatory increase in gonadotropins was documented. In fact, LH and FSH decreased over the same period, suggesting that the mechanism of toxicity must include a central component that involves either the hypothalamus and/or the pituitary gland. Taken together, these data suggest either that, in the presence of pre-existing gonadal damage, even a minor reduction in gonadotropin drive can decrease T levels, and/or that a direct gonadal effect of the crizotinib also is occurring.

Additional insights can be gained by investigating FSH, LH, and PRL levels both in the original CTG and in the comparator low-T NCTG (Tables 1, 2). First, in the NCTG, although the upper limit of the observed range indicates that FSH and LH levels may be elevated in some individuals, the mean FSH and LH levels are not above the upper limits of normal for either gonadotropin. This suggests that patients who have cancer with low T indeed may manifest low T at FSH and LH levels that still are within the normal range because of a requirement for increased basal gonadotropin drive.17, 22 Second, in the CTG, despite prolonged crizotinib therapy, mean FSH levels were not reduced but, in fact, were at the top of the normal range; and neither FSH levels nor LH levels were statistically significantly different from those in the NCTG. This is in contrast to the immediate reductions in gonadotropin levels noted in the changes illustrated for 2 patients in Figure 2 and raises some doubt about whether a central effect alone is responsible for low T in crizotinib-treated patients. Because, in some crizotinib-treated patients, FSH and LH levels can exceed the upper limit of normal (as demonstrated by the upper end of the observed range for each), we hypothesize that an additional, direct gonadal effect of crizotinib probably exists.

Crizotinib is primarily an ALK and cMET inhibitor, although other kinases also may be affected, depending on drug exposure.23 Expression of both ALK and cMET has been described in the testes.24, 25 Expression of both ALK and cMET also occurs throughout the central nervous system and has been described in some aspects of the hypothalamic-pituitary axis.24-30 Assuming that action on either ALK, cMET, or both is responsible for the observed low T with concomitant reduction in gonadotropin levels, these expression patterns also are consistent with the potential for both central and direct gonadal effects of crizotinib.

Preclinical studies using kinase-specific inhibitors are required to accurately determine whether the effect of crizotinib is related to the inhibition of cMET, ALK, or other receptor tyrosine kinase targets and the exact site(s) of action in the hypothalamic-pituitary-testicular axis. Such studies also offer the potential for providing unique insights into the molecular signaling pathways involved in aspects of normal human endocrine physiology. If this is entirely an ALK-specific effect, then the potential for using T changes as a pharmacodynamic marker during the early phase development of other ALK inhibitors should be considered.31 In contrast, if it reflects a combination of both ALK and MET or other off-target kinase inhibition, then hormone side effects offer the potential to differentiate crizotinib from other more specific ALK inhibitors currently in development.32

Because the detection of low T was noted after long-term use of crizotinib in the majority of patients, we were unable to correlate the timing of the decrease in T with the recorded side effects of therapy. Certainly some of the side effects associated with low T (notably, fatigue) do occur with crizotinib. In the index crizotinib-treated case, the patient reported the development of persistent fatigue after only 2 cycles of therapy despite responding to the study drug, consistent with the timing of changes in T levels observed in our dynamic series (Fig. 2). Many symptoms of hypogonadism specific to low T, such as sexual disinterest, erectile dysfunction, decrease in facial hair growth, loss of muscle mass, and increase in fat mass, may have been missed by study investigators who were not investigating specifically for these features. To our knowledge, objective measurements, such as bone mineral density and muscle mass, which can alter with hypogonadism, have not been assessed in any of the crizotinib studies to date.

Although the effect of crizotinib on T was reversible on drug discontinuation, patients may remain on crizotinib for many months.10 The potential long-term consequences of undetected and untreated low T on crizotinib in this population are unknown. Low T in cancer patients has been correlated with reduced QoL, fatigue, and worsened survival.12, 33 Total T and free T levels decline with age,21, 34 and cross-sectional surveys have reported a prevalence of hypogonadism, defined primarily by low T, of approximately 20% to 40% in men aged >40 years.20, 35, 36 Studies in patients with metastatic cancer have reported a higher prevalence in the range of 40% to 70% based on T levels,12, 13 possibly related to malnutrition, opioid use, or prior cytotoxic chemotherapy causing direct gonadal damage.13-17 Low T in cancer patients has been correlated with fatigue, decreased QoL, sexual disinterest, and inferior survival.12, 33 In addition, as T levels fall with age, some older patients who have received crizotinib already may be manifesting hypogonadism and may not notice drug-induced changes in their T levels. In contrast, the greater frequency of ALK gene rearrangements in younger patients may render a significant proportion of the crizotinib-treated population particularly susceptible to experiencing the side effects associated with drug-induced low T.32 In a noncancer population, extensive literature has documented that low T levels lead to sexual dysfunction, depressive symptoms, and fatigue as well as decreased muscle mass, decreased bone density. and increased cardiovascular risk.37-40

Physiologic T treatment, in the form of transdermal patches or gel, can normalize T levels, but the specific clinical benefits continue to be investigated. Two small trials that evaluated T replacement in cancer patients reported improved fatigue, reduced weight loss, and increased physical activity but were unable to report improved QoL.41, 42 However, both of those trials included significant numbers of patients who had normal T levels, so the effect of treatment in cancer patients with low T is uncertain. Therapy in noncancer patients leads to increased bone density, improved muscle mass, increased hematocrit, and improved libido while potentially worsening symptoms of benign prostatic hypertrophy and sleep apnea.43-46 Although, anecdotally, our patients who were receiving T treatment often reported reduced fatigue, improved cognition, increased libido, and increased exercise tolerance (data not shown), a prospective clinical trial will be required to formally assess whether T replacement in crizotinib-treated patients with NSCLC leads to benefits and toxicities similar to those observed in other cancer patients.

Whether crizotinib may have comparable hypothalamic-pituitary-gonadal effects in women has not yet been investigated. Cyclical effects in premenopausal women and different reference ranges in postmenopausal women make any initial assessments more complicated, but these also could be addressed within a prospective study.

In summary, low T levels were documented in all of our male patients with metastatic NSCLC who received crizotinib treatment. Changes appeared to be both rapid and reversible on commencement and cessation of dosing, respectively, confirming a direct causal relation with crizotinib. Given the observed decrease in FSH and LH levels with commencement of therapy in several patients, the site of action must include a central component, but additional, direct testicular effects cannot be excluded. In vitro data will be required to fully understand the mechanism of these effects on the hypothalamic-pituitary-testicular axis. We suggest that clinicians should monitor T levels in all men who receive crizotinib and should refer those with low T for discussion of the risks and benefits of T therapy.


No specific funding was disclosed.


Drs. Weickhardt, Doebele, and Camidge have received speaking honoraria from Pfizer.