Gender impact on safety and efficacy in lenvatinib treated patients with radioiodine-refractory differentiated thyroid cancer (GISEL study)

Abstract

Introduction

Little is known about sex differences in lenvatinib treatment safety and efficacy.

Methods

Real-word retrospective Italian multicenter study enrolling patients with radioiodine-refractory differentiated thyroid cancer treated with lenvatinib.

Results

A total of 138 patients (64 females) were included, with a median follow-up of 26 months (2–72). More men performed physical activities (34% vs 17%, P = 0.024). The frequency of smoking and alcohol consumption was higher in men (58% vs 33%, P = 0.003; 45% vs 17%, P = 0.001). We did not find sex differences in lenvatinib dose reduction due to adverse events (AEs) (78% females vs 85% males). Ninety-nine percent of patients developed at least one adverse event (AE), with no sex difference in their number and the time to first AE. Severe AEs occurred in 74% of males and 66% of females (P = 0.398), with a mean dose of 18.2 mg (±5.7), and a median time to the first serious AE of 9 weeks (1–154). Stomatitis/mucositis and hematological disorders were more frequent in females (48% vs 30%, P = 0.016; 17% vs 4%, P = 0.011). Gastrointestinal disorders were higher in males (15% vs 2%, P = 0.010). Eighty-seven patients interrupted lenvatinib due to AEs (median time: 3 months (0–48), mean dose: 17 mg ±5.5). Discontinuation occurred in 21 patients, five for severe AEs. No sex differences were found in progression-free survival, overall survival or disease control rate. Liver metastases were associated with disease progression (HR: 3.73, 95% CI: 1.06–13.12, P = 0.040) or death (HR: 4.82, 95% CI: 1.75–13.25, P = 0.002) only in females.

Conclusion

Lenvatinib is effective in both sexes and exhibits a good safety profile, with a sex difference in the frequencies of some adverse events.

Introduction

Lenvatinib (LEN) is a multi-target tyrosine kinase inhibitor (TKI) able to inhibit vascular endothelial growth factor receptors (VEGFRs) types 1, 2 and 3; fibroblast growth factor receptors (FGFRs) types 1–4; platelet-derived growth factor receptors (PDGFRs) α, β and KIT. LEN exerts its action through the inhibition of angiogenesis (mostly mediated by VEGFR blocking) and tumor cell migration and invasion (suppressing FGFR and PDGFR) (1).

LEN has been investigated in radioiodine-refractory (RR) progressive, advanced or metastatic differentiated thyroid cancers (DTCs), in the phase 3 randomized controlled trial SELECT, which led to the approval of the drug in the USA, Europe and Japan (2). In this study, LEN (24 mg daily) significantly increased progression-free survival (PFS) compared to placebo (18.3 vs 3.6 months; risk for death or progression 0.21, 99% CI: 0.14–0.31, P < 0.001), with a remarkable response rate over placebo (65 vs 2%) also reducing the tumor burden, with a median maximum percentage change in tumor size of −42.9%). Although the efficacy is remarkable, side effects are considerable and not all patients have the same risk-to-benefit ratio from LEN treatment. The efficacy of LEN has been confirmed in specific populations of patients, such as the elderly, defined as patients with age over 65 years (approximately 40% of subjects included in the study) (3). The results from the SELECT study have not been disaggregated by sex, and it is only possible to affirm that, in a subgroup analysis, LEN increased PFS both in males and in females (2), and safety according to sex has not been evaluated. In addition, a Chinese clinical trial on the pharmacokinetics of lenvatinib demonstrated higher area under the curve and maximum concentration in females than in males (4).

Results derived from controlled clinical trials are rarely replicable in real life because these trials are usually conducted in selected populations in a highly controlled setting optimized to show the effect of the drug (5). In addition, there is growing evidence that sex plays a role not only in the incidence and progression of neoplasms but also in the response to therapy. Sex-related differences in toxicities and treatment response of some antineoplastic drugs have been demonstrated, and this is caused by differences in pharmacokinetic and pharmacodynamic characteristics that distinguish males (M) and females (F) (such as body composition, enzymatic activity and fat distribution) but also by differences in immune response (6). In addition, females had a higher rate of side effects (7). Nowadays, few real-life studies with LEN in RAI-refractory DTC have been published (8, 9, 10, 11, 12, 13, 14, 15, 16, 17) and, to our knowledge, no data exist about sex and gender differences in its efficacy and safety. This multicentric study aimed to evaluate sex differences in the efficacy and toxicity of LEN in a real-life setting of patients affected by RR-DTC, as recommended by the consensus of the Italian Association of Medical Oncology on gender oncology (18).

Materials and methods

Study design

This was a retrospective national multicenter observational study, performed in eight Italian centers with high expertise in the management of advanced thyroid cancer, registered on clinicaltrial.gov (NCT05789667). The study enrolled patients aged >18 years affected by progressive RR-DTC treated with LEN. RR-DTC was defined as: metastatic disease that does not take up radioactive iodine at the time of initial treatment; tumors that lose the ability to take up radioactive iodine after previous evidence of uptake; radioactive iodine uptake only in some lesions or metastatic disease that progresses despite substantial uptake of radioactive iodine (19).

The decision regarding the start of LEN treatment and the dose was made by a multidisciplinary team of experts in each institutional thyroid tumor board based on tumor burden, tumor-related symptoms, size and localization of metastases and patient characteristics such as age and comorbidities, according to guidelines (20, 21). In particular, LEN treatment was started in case of progressive disease with symptomatic, multiple lesions (single lesions are treated by locoregional therapy). Comorbidities and patients’ ECOG examination were used for choosing the starting dose, preferring the maximum dose (24 mg daily) whenever possible but starting from a lower dose in case of a higher risk of side effects (such as the risk of fistula or low body weight) or the presence of comorbidities. Adverse events were graded according to CTCAE version 5.0 (22). As this was an observational study, each physician or tumor board decided how to manage the adverse events, including dose reduction or treatment interruption. In most cases, LEN therapy was interrupted (temporarily) or discontinued (permanently) after grade 3 or higher adverse events.

All patients gave written informed consent to participate. The study was conducted under the approval of the local ethics committee of Regina Elena National Cancer Institute (reference number: 1316/20). This study was conducted in agreement with the National Law and the Declaration of Helsinki. This study followed the STROBE guidelines for observational studies (23) and SAGER guidelines (Sex and Gender Equity in Research) (24).

Data collection

All patients included in the study underwent periodic clinical, radiological and biochemical evaluation as recommended by good clinical practice. The evaluation of LEN adverse events was performed during the periodic clinical evaluations in the outpatient clinic using the last available CTCAE edition for assessing grade. Careful reporting of all adverse events and their attribution to LEN treatment is part of the clinical practice of the participating centers because it is essential for patients’ management and the decision on dose reduction, interruption or discontinuation. For the study, a standardized data collection form was used for collecting data (using medical reports as source documents), and adverse events were reclassified according to CTCAE version 5.0 if necessary (22). The following data were collected for each patient: sex; age at diagnosis and age at the start of LEN treatment; menopausal status (for women); ethnicity; nationality; distance from the hospital; gender determinants of health (including levels of education, marital status, number of children, caregiver role, height, weight, body mass index (BMI), cigarette smoking, alcohol intake – considered positive in case of at least seven alcohol units weekly – and physical activity); data on the past medical history including comorbidities and concomitant medication; data on the thyroid cancer (histology, staging at diagnosis, localization of secondary lesions, treatments performed including surgery, radioiodine treatment and other TKIs); data about dose and time of initiation of LEN; data on all adverse events including the type, the grade, the time of onset, the eventual treatment schedule modification and the recovery of the adverse events; and data on treatment efficacy and survival.

Safety and efficacy outcomes

The primary objective of the study was the sex-stratified assessment of the safety of LEN. Adverse events have been classified as follows: arterial hypertension (defined as blood pressure >140/90 mmHg); QT prolongation (>480 msec); cardiological disease (all cardiological alterations excluding QT prolongation); proteinuria; stomatitis and mucositis; hand and foot syndrome; nausea and anorexia; weight loss (>5% from baseline); gastrointestinal disorders (defined as diarrhea and abdominal pain); electrolyte disorders; increased TSH; hematological disorders (alteration in platelet count, anemia and leucopenia); fatigue; headache; cholecystitis/cholelithiasis; fistula; hemorrhage; musculoskeletal pain; and renal failure. The grading of adverse events followed CTCAE version 5.0 (22), considering grades 3 and 4 as severe adverse events.

The secondary objective was the evaluation of the treatment response. Efficacy was evaluated according to response evaluation criteria in solid tumors (RECIST version 1.1). We evaluated the best response, defined as the best response to treatment (at any time point of evaluation). The overall response rate (ORR) was defined as the proportion of patients with partial or complete response, and the disease control rate (DCR) was assessed as the proportion of patients with stable disease, partial or complete response. Median PFS and median overall survival (OS) were also calculated as efficacy outcomes.

Statistical analysis

Variables of interest were expressed as frequencies and percentage values, while continuous variables were expressed as mean ± standard deviation or median and minimum–maximum range, as appropriate. Descriptive statistics were calculated for all variables of interest. Differences between sexes and genders were tested with Pearson’s chi-square non-parametric test or the most suitable test for continuous variables. Duration of response, OS and PFS analysis were conducted through the use of the Kaplan–Meier method. The log-rank test was applied to assess potential differences between subgroups. A P-value <0.05 was considered statistically significant. All statistical analyses were performed using Statistical Package for the Social Sciences (SPSS version 21.0).

Sample size calculation

The primary objective of the study was to evaluate the safety and toxicity profile in terms of dose reduction based on sex. Fixing the statistical power at 80%, the significance level at 0.05 and the percentage of M who could experiment dose reduction equal to 59% (2), with a sample size of 50 patients (25 M and 25 F), we estimated, through the use of Fisher’s exact test, an expected percentage of F who could experiment a dose reduction equal to 94%.

Results

Patients’ characteristics

One hundred thirty-eight patients were enrolled in this study: 64 were F (46%) and 74 M (54%), with a median follow-up of 26 months (2–72). The total median age at surgery (corresponding to the age at thyroid cancer diagnosis) was 61 years (22–85); the median time between surgery and diagnosis of radioiodine refractoriness was 3 years (0–30). The total radioiodine therapy dose was 300 mCi (30–1,750). The median age at the start of LEN was 69 years (range min–max: 38–96), with a mean latency time of 7.9 years (±7.4) between surgery and LEN treatment. Among the 138 patients, the most frequent histology was papillary thyroid cancer, followed by follicular (27%), poorly differentiated (22%) and oncocytic carcinoma (8%), without sex difference. Ten percent of patients received treatment with other TKIs before LEN. At the time of LEN starting, 81% of patients had multiple metastatic lesions; the most frequent metastatic site at treatment starting was the lung (including pleura and/or mediastinum), followed by local relapse and bone. Most patients had comorbidities at the start of LEN treatment (median 1, 0–6). The most frequent were hypertension (64% of patients), dyslipidemia (20%), cardiomyopathy (13%) and other malignancies (13%). No sex difference was found for all these parameters. All the patients’ characteristics are summarized in Table 1.

Table 1.

Patients’ characteristics and gender-related determinants of health. Continuous data are expressed as median (min–max range) or mean ± SD as appropriate, according to variable distribution. Other data are presented as n (%). In case of missing data, percentages are calculated not on the total number of patients but only on the patients for whom data are available.

	Overall	Males	Females	P-value
Patients, n (%)	138 (100)	74 (54)	64 (46)
Age at surgery, years	61 (22–85)	61 (22–81)	59 (33–85)	0.545
Age at lenvatinib initiation, years	69 (38–96)	69 (38–85)	70 (45–96)
Time between surgery and RRD	3 (0–30)	3 (0–30)	4 (0–24)	0.408
Time between surgery and lenvatinib starting	7.9 ± 7.4	7.3 ± 6.7	8.7 ± 8.2	0.460
Staging at diagnosis, n	131	71	60	0.209
I	33 (25%)	16 (22%)	17 (28%)
II	41 (31%)	18 (25%)	23 (38%)
III	6 (5%)	4 (6%)	2 (4%)
IV A	2 (2%)	2 (3%)	0
IV B + C	49 (37%)	31 (44%)	18 (30%)
Histology, n	136	73	63	0.921
Papillary	58 (43%)	33 (45%)	25 (40%)
Follicular	37 (27%)	19 (26%)	18 (28%)
Oncocytic carcinoma	11 (8%)	6 (8%)	5 (8%)
Poorly differentiated	30 (22%)	15 (21%)	15 (24%)
Radiometabolic therapy dose, mCi	300 (30–1,750)	300 (30–1,200)	300 (80–1,750)	0.427
Previous TKI treatment	14 (10%)	10 (14%)	4 (6%)	0.167
Metastasis site
Lung (or pleura or mediastinum)	122 (88%)	68 (92%)	54 (84%)	0.169
Bone	57 (41%)	36 (49%)	21 (33%)	0.060
Local relapse	67 (49%)	36 (49%)	31 (48%)	0.980
Brain	9 (7%)	4 (5%)	5 (8%)	0.568
Liver	11 (8%)	6 (8%)	5 (8%)	0.949
Adrenal	5 (4%)	4 (5%)	1 (2%)	0.228
ECOG at treatment starting				0.157
0	85 (62%)	51 (69%)	34 (53%)
1	38 (28%)	16 (22%)	22 (34%)
2+	15 (11%)	7 (9%)	8 (13%)
BMI at treatment starting	26.5 (14.1–47)	26.8 (18.1–47)	25.4 (14.1–45.7)	0.101
No. of relevant comorbidities	1 (0–6)	2 (0–5)	1 (0–6)	0.494
Type of relevant comorbidities
Hypertension	88 (64%)	51 (69%)	37 (58%)	0.176
Dyslipidemia	28 (20%)	14 (19%)	14 (22%)	0.667
Cardiomyopathy	18 (13%)	12 (16%)	6 (9%)	0.234
Other cancer	18 (13%)	11 (15%)	7 (11%)	0.495
Diabetes mellitus	12 (9%)	4 (5%)	8 (13%)	0.225
Renal failure	7 (5%)	5 (7%)	2 (3%)	0.450
Sarcopenia	6 (4%)	1 (1%)	5 (8%)	0.096
Hepatic failure	2 (1%)	1 (1%)	1 (2%)	1.000
Gender related determinants of health
Menopausal status – yes	-	-	57 (89%)
Distance from referring hospital				0.420
<10 km	25 (18%)	15 (20%)	10 (16%)
10–50 Km	29 (21%)	14 (19%)	15 (23%)
>50 Km (in the same region)	35 (25%)	22 (30%)	13 (20%)
>50 Km (other region)	49 (36%)	23 (31%)	26 (41%)
Ethnicity – Caucasian	138 (100%)	74 (100%)	64 (100%)	n.e.
Nationality				0.213
Italian	136 (99%)	74 (100%)	62 (97%)
Other	2 (1%)	0	2 (3%)
Role as caregiver – yes	9 (7%)	2 (3%)	7 (11%)	0.081
Children, n	132	68	64	0.213
Yes	110 (83%)	54 (79%)	56 (88%)
Marital status, n	136	72	64	0.270
Single	8 (6%)	5 (7%)	3 (5%)
Married	109 (80%)	60 (83%)	49 (77%)
Widower	14 (10%)	4 (6%)	10 (16%)
Divorced	5 (4%)	3 (4%)	2 (3%)
Level of education, n	92	41	51	0.420
Primary school	27 (29%)	9 (22%)	18 (35%)
Middle school	20 (22%)	9 (22%)	11 (22%)
High school	28 (30%)	13 (32%)	15 (29%)
Degree	17 (19%)	10 (24%)	7 (14%)
Physical activity – yes	36 (26%)	25 (34%)	11 (17%)	0.024
Smoking status, n	136	72	64	0.003
Yes	13 (9%)	6 (8%)	7 (11%)
Ex	50 (37%)	36 (50%)	14 (22%)
No	73 (54%)	30 (42%)	43 (67%)
Alcohol intake, n	121	62	59	0.001
Yes	38 (31%)	28 (45%)	10 (17%)

Ne, not evaluable; RRD, radioiodine-refractory diagnosis; TKI, tyrosine kinase inhibitor; BMI, body mass index. P values in bold are statistically significant.

Gender-related determinants of health

We found some gender differences in the main determinants of health. Considering lifestyle parameters, a higher percentage of men performed physical activities (34 vs 17%, P = 0.024), but a higher proportion of men were smokers or ex-smokers compared to women (58 vs 33%, P = 0.003) and alcohol consumption was higher in men than in women (45 vs 17%, P = 0.001). Many patients traveled long distances to reach the referral hospital (61% of patients more than 50 km), without gender difference. These data highlight the availability of patients to reach referral centers, even if this implies moving far away from their houses with monthly/bimonthly frequency of visits.

Regarding socio-economic variables, no difference in educational level, marital status, number of children and the role of caregiver for another person was found according to gender. Gender-related determinants of health are summarized in Table 1.

We therefore correlated the gender-related determinants of health with LEN safety and efficacy outcomes. Regarding safety, stratifying patients according to physical activities, in the subgroup of sedentary behavior, we found that women had a shorter time to first severe adverse events (4 months, 0–114 vs 10, 0–102, P = 0.021), with a lower dose (14 mg, 2–24 vs 20, 1–24, P = 0.047). Overall, sedentary F had a higher mean number of total adverse events compared to sedentary M (5.5 ± 2.8 vs 4.5 ± 2.3, P = 0.04). Smoking and alcohol consumption did not impact the total number of adverse events, the number of severe adverse events or the development of more common adverse events including arterial hypertension, fatigue, gastrointestinal disorders or stomatitis/mucositis.

Regarding efficacy, we found no impact of physical activities, smoking status, alcohol intake or BMI on OS and PFS in both genders (see also Table 5).

Table 5.

OS and progression-free survival analysis: Cox regression model results. Multivariate analysis has not been performed for the low number of significant predictors.

Parameter	Comparison	Overall			Males			Females
		HR	95% CI	P	HR	95% CI	P	HR	95% CI	P
OS
Age at start of lenvatinib	Continuous	1.02	0.99–1.05	0.121	1.03	0.99–1.07	0.133	1.01	0.97–1.06	0.509
Basal ECOG	1+ vs 0	1.35	0.77–2.36	0.290	1.57	0.73–3.36	0.244	1.22	0.53–2.82	0.644
Previous TKI	Yes vs no	1.69	0.76–3.79	0.200	1.91	0.71–5.12	0.200	1.45	0.34–6.24	0.617
Bone metastasis	Yes vs no	0.75	0.42–1.33	0.322	0.89	0.42–1.86	0.751	0.56	0.22–1.44	0.227
Liver metastasis	Yes vs no	1.63	0.72–3.70	0.241	0.53	0.12–2.29	0.393	4.82	1.75–13.25	0.002
Lung metastasis	Yes vs no	1.19	0.51–2.79	0.693	24.48	0.13–4,718.23	0.234	0.64	0.25–1.65	0.361
Physical activity	Yes vs no	0.99	0.52–1.90	0.974	0.99	0.45–2.22	0.978	0.95	0.28–3.23	0.935
Smoking status	Yes vs no + former	1.01	0.42–2.43	0.989	0.99	0.32–3.04	0.986	0.89	0.21–3.84	0.881
Alcohol intake	Yes vs no	1.18	0.62–2.25	0.623	2.08	0.87–4.99	0.102	0.45	0.11–1.96	0.290
BMI	Continuous	0.98	0.93–1.03	0.485	0.94	0.87–1.02	0.133	1.02	0.95–1.10	0.598
Histotype	PD vs other	0.52	0.23–1.16	0.109	0.41	0.12–1.37	0.148	0.68	0.23–2.01	0.482
Sex	F vs M	0.99	0.56–1.74	0.968
Progression free survival
Age at start of lenvatinib	Continuous				1.01	0.98–1.05	0.473	0.97	0.93–1.02	0.238
Basal ECOG	1+ vs 0	1.24	0.73–2.13	0.429	1.11	0.53–2.33	0.790	1.76	0.73–4.28	0.210
Previous TKI	Yes vs no	0.95	0.38–2.39	0.910	1.23	0.47–3.21	0.676	0.45	0–153.34	0.454
Bone metastasis	Yes vs no	1.30	0.77–2.20	0.333	1.02	0.53–2.00	0.946	1.74	0.74–4.11	0.208
Liver metastasis	Yes vs no	1.56	0.70–3.45	0.276	1.01	0.35–2.89	0.990	3.73	1.06–13.12	0.040
Lung metastasis	Yes vs no	0.99	0.45–2.20	0.985	2.03	0.48–8.52	0.335	0.51	0.18–1.43	0.202
Physical activity	Yes vs no	0.97	0.52–1.81	0.913	1.04	0.50–2.16	0.922	0.60	0.14–2.60	0.497
Smoking status	Yes vs no + former	1.40	0.63–3.13	0.412	1.08	0.37–3.20	0.886	2.26	0.64–7.98	0.207
Alcohol intake	Yes vs no	1.79	0.99–3.24	0.056	1.92	0.89–4.14	0.095	1.27	0.41–3.90	0.680
BMI	Continuous	0.97	0.92–1.02	0.210	0.94	0.86–1.01	0.090	0.99	0.91–1.08	0.826
Histotype	PD vs other	1.04	0.56–1.94	0.893	0.86	0.37–2.01	0.723	1.41	0.57–3.53	0.461
Sex	F vs M	0.73	0.42–1.26	0.254

TKI, tyrosine kinase inhibitors; BMI, body mass index; F, females; M, males; PD, poorly differentiated; OS, overall survival. P values in bold are statistically significant.

Treatment safety

The initial treatment dose was the standard dose of 24 mg for 48% of patients, while 17% received 20 mg as the starting dose and 35% of patients started with a dose lower than 20 mg, without significant sex differences. The starting dose was negatively correlated with the age at the start of LEN treatment (R²: 0.386, P < 0.001); coherently, the median age was significantly different between the groups of patients treated with 24 mg (66 years, 43–84), 20 mg (70, 48–85) and 14 mg (76, 58–89), P < 0.001. No correlation was found between the starting dose and the total number of comorbidities.

The primary endpoint of the study was the evaluation of LEN toxicities in terms of dose reduction. We found that 78% of F and 85% of M reduced LEN dose due to adverse events, without significant difference (P = 0.286). One hundred thirty-six of 138 (99%) patients developed at least one adverse event, without sex differences (100% of M, 97% of F). Only two F did not present any adverse events while all M presented at least one adverse event. No significant sex differences were found in the whole number of adverse events (n = 7 in M and n = 8 in F, P = 0.268). The mean time to the first adverse event was 5.5 weeks (±10.4) without sex or age difference. Eighty-two percent of patients (113) reduced the dose of LEN after a median of 3 months (0–34), without sex difference.

The two most common AE were hypertension, reported in 73% of patients (69% of M and 77% of F), and fatigue, reported in 65% of patients (61% M and 67% F). These two side effects were followed by, for M, gastrointestinal disorders (58%), nausea and anorexia (54%), and for F, by nausea and anorexia (58%) and weight loss (55%).

Regarding the single type of adverse events, sex differences were observed in the frequency of stomatitis and mucositis, observed more frequently in F (48 vs 30%, P = 0.016) and hematological disorders, reported in 17% of F versus 4% of M (P = 0.011). Data about adverse events are reported in Table 2 and Fig. 1 summarizes sex differences in the type of adverse events.

Table 2.

Number and type of all adverse events according to sex. Data are presented as n (%), median (range min–max) or as mean ± SD.

	Overall	Males	Females	P-values
Lenvatinib dose (at starting)				0.053
24 mg	66 (48)	40 (54)	26 (41)
20 mg	24 (17)	15 (20)	9 (14)
<20 mg	48 (35)	19 (26)	29 (45)
Patients with at least one AE	136 (99)	74 (100)	62 (97)	0.213
Time of first AE, weeks	5.5 ± 10.4	6.93 ± 12.4	3.82 ± 7.1	0.226
Total number of adverse events during lenvatinib treatment*	7 (0–20)	7 (1–19)	8 (0–20)	0.268
Hypertension	100 (73)	51 (69)	49 (77)	0.183
QT prolongation	4 (3)	3 (4)	1 (2)	0.625
Renal failure	8 (6)	5 (7)	3 (5)	0.727
Proteinuria	37 (27)	16 (22)	21 (33)	0.110
Stomatitis and mucositis	53 (39)	22 (30)	31 (48)	0.016
Hand-foot syndrome	51 (37)	25 (34)	26 (41)	0.328
Nausea and anorexia	77 (57)	40 (54)	37 (58)	0.510
Weight loss	67 (49)	32 (43)	35 (55)	0.125
Gastrointestinal disorders	75 (54)	43 (58)	32 (50)	0.448
Electrolyte disorders	4 (3)	1 (1)	3 (5)	0.231
TSH increased	37 (27)	17 (23)	20 (31)	0.226
Hematological disorders	14 (10)	3 (4)	11 (17)	0.011
Fatigue	88 (65)	45 (61)	43 (67)	0.299
Headache	2 (1)	1 (1)	1 (2)	1.000
Cholecystitis/Cholelithiasis	19 (14)	11 (15)	8 (13)	0.742
Fistula	4 (3)	2 (3)	2 (3)	1.000
Hemorrhage	9 (7)	7 (9)	2 (3)	0.181
Musculoskeletal pain	20 (15)	12 (16)	8 (13)	0.587
Cardiological diseases	5 (4)	3 (4)	2 (3)	1.000
Mixed*	43 (32)	23 (31)	20 (31)	0.883
Time at first interruption, months	3 (0–48)	4 (0–48)	3 (0–45)	0.942
Dose at first interruption, mg	17.0 ± 5.5	17.5 ± 5.7	16.1 ± 5.4	0.169
Time at first dose reduction, months	3 (0–34)	4 (0–25)	2.5 (0–34)	0.202
Time at discontinuation, months	15 (2–37)	23 (7–32)	14 (2–37)	0.548

^*Mixed adverse events include epilepsy, chest pain, dysphagia, nosebleeds, insomnia and pleural effusion.

P values in bold are statistically significant.

Figure 1.

Tornado plot showing the frequencies of the main adverse events in males and females.

No sex differences were found, neither in the percentage of patients developing grade ≥3 adverse events (74% of M and 66% of F, P = 0.398), nor in the mean LEN dose at the first occurrence of a grade ≥3 adverse event (18.2 mg ± 5.7). The median dose was lower in the subgroup of patients aged >70 years than in patients aged <70 years (14 mg, 4–24 vs 24 mg, 4–24, P < 0.001). The median time at the first severe adverse event was 9 weeks (range min 1–154), without sex or age difference. No sex difference was found in the type of adverse events reported, except for gastrointestinal disorders (15% of M and 2% of F, P = 0.010), as summarized in Table 3.

Table 3.

Severe adverse events (grade 3 and 4), overall and stratified by gender. Data are presented as n (%), mean ± SD or as median (range).

	Overall	Males	Females	P-value*
Patients with at least one AE severe (≥G3)	97 (70%)	55 (74%)	42 (66%)	0.398
Dose at the first G3 adverse events	18.2 ± 5.7	19.0 ± 5.4	17.1 ± 5.9	0.174
Time at the first G3 adverse events, weeks	9 (1–154)	10 (1–154)	7 (1–114)	0.574
Type of adverse events
Adverse event/Grade
Hypertension
G3	44 (32%)	21 (28%)	23 (36%)	0.279
QT prolongation
G3	3 (2%)	2 (3%)	1 (2%)	1.000
Renal failure
G3 + G4	4 (3%)	2 (3%)	2 (3%)	1.000
G3	3 (2%)
G4	1 (1%)
Proteinuria
G3	9 (7%)	4 (5%)	5 (8%)	0.732
Stomatitis and mucositis
G3	10 (7%)	4 (5%)	6 (9%)	0.512
Hand-foot syndrome
G3	3 (2%)	2 (3%)	1 (2%)	1.000
Nausea and anorexia
G3	13 (10%)	6 (8%)	7 (11%)	0.570
Weight loss
G3	21 (15%)	12 (16%)	9 (14%)	0.785
Gastrointestinal disorders
G3	12 (16%)	11 (15%)	1 (2%)	0.010
Electrolyte disorders
G3 + G4	2 (2%)	0	2 (3%)	0.206
G3	1 (1%)
G4	1 (1%)
Hematological disorders
G3	1 (1%)	1 (1%)	0	1.000
Fatigue
G3	17 (12%)	10 (14%)	7 (11%)	0.696
Cholecystitis/Cholelithiasis
G3	12 (9%)	9 (12%)	3 (5%)	0.224
Fistula
G3 + G4	3 (2%)	1 (1%)	2 (3%)	0.592
G3	2 (1%)
G4	1 (1%)
Musculoskeletal pain
G3	4 (3%)	3 (4%)	1 (2%)	0.625
Cardiac disorders
G3	3 (2%)	2 (3%)	1 (2%)	1.000
Other
G3 + G4	9 (7%)	4 (5%)	5 (8%)	0.732
G3	5 (4%)
G4	4 (3%)

^*Chi-square non parametric test.

G3, grade 3; G4, grade 4. P values in bold are statiistically significant.

Eighty-seven patients interrupted LEN due to adverse events (49 M and 38 F); 48% of these patients had already reduced the initial LEN dose while the remaining were taking the initial treatment dose. The median time at first interruption of LEN was 3 months (range min–max 0–48 months) and the overall dose was 17 mg (±5.5), without sex difference (Table 3).

In 21 patients (15%), LEN was discontinued in five patients for severe adverse events, three for new diagnoses which contraindicated treatment with LEN (myocardial infarction or other malignancies), one for downstaging of the disease (result of good treatment response), nine for progressions of disease and three for ECOG worsening. The median time to discontinuation was 15 months (range min 2-max 35) without sex difference (23 months in M and 14 months in F).

A supplementary analysis on the most relevant grade 3–4 adverse events has been performed (including fatigue, hypertension, weight loss, renal failure and proteinuria). For the main frequent and relevant grade 3–4 adverse events, we assessed the time of onset, the dose at which severe adverse events occurred and the physician’s decision regarding discontinuation or dose reduction (as reported in Supplementary Material (see section on Supplementary materials given at the end of the article)). Briefly, hypertension occurred earlier than other adverse events (3 weeks), followed by proteinuria (12 weeks), gastrointestinal disorders and fatigue (approximately 20 weeks), weight loss (41 weeks) and renal failure (59 weeks). Finally, physicians were more prone to interrupt treatment in cases of renal failure and proteinuria, while hypertension and fatigue were mostly managed with additional treatments and treatment dose reduction.

Efficacy

Considering efficacy outcome, the ORR of LEN was 36% of patients and DCR was 91%. No sex differences were found in terms of ORR and DCR, OS and time to death, as summarized in Table 4. Median PFS was 44 months (28.0–60.0), without sex differences (see Supplementary Fig. 1). We also evaluated the role of some possible risk factors on OS and PFS according to sex. The presence of liver metastases was associated with disease progression (HR: 3.73, 95% CI: 1.06–13.12, P = 0.040) or death (HR: 4.82, 95% CI: 1.75–13.25, P = 0.002) only in F, as summarized in Table 5.

Table 4.

Efficacy evaluation. Data are presented as median (range min–max) or as n (%).

	Overall	Males	Females	P-values
Median FU, months	26 (2–72)	26 (2–72)	26 (2–61)	0.590
Best response, n = 135				0.239
Partial response	48 (36)	23 (32)	25 (40)
Stable disease	75 (55)	41 (56)	34 (55)
Progressive disease	12 (9)	9 (12)	3 (5)
Overall response	48/135 (36%)	23 (32)	25 (40)	0.286
DCR	123/135 (91%)	64 (88)	59 (95)	0.128
OS status, n = 137				0.518
Alive	86 (63)	44 (60)	42 (66)
Death	51 (37)	29 (40)	22 (34)
PFS, months
Only progression	44 (28.0–60.0)	42 (20.6–63.4)	53 (34.1–71.9)	0.249
Progression + death	29 (16.9–41.9)	26 (11.9–40.1)	29 (9.8–48.2)	0.867
Time to death, months*
Overall	56 (41.4–70.9)	53 (33.4–72.6)	NE	0.968
Thyroid cancer	56 (44.4–67.6)	53 (33.4–72.6)	NE	0.591

FU, follow-up; CI, confidence interval; NE, not estimable; DCR, disease control rate; OS, overall survival; PFS, progression-free survival.

^*Data are the median (95% CI).

Discussion

This is the first real-world study aimed at evaluating sex and gender differences in the safety and efficacy of LEN. At the moment, no data disaggregated by sex are available for LEN treatment in patients affected by RR-DTC. The importance of considering sex and gender differences in oncology has been widely recognized, as it affects many aspects of tumors such as pathophysiology, clinical manifestations and response to treatment (25). The term gender includes not only the biological difference between F and M but also socio-cultural and economic aspects (26). Unlike other non-reproductive cancers that show a male predominance, thyroid cancer is more common in F (27). In patients younger than 55 years of age, thyroid cancers in M appear to have a worse prognosis than in F (28), even if this result can be overestimated by the higher incidence of other prognostic factors such as an increase of TERT promoter mutations (29) or the diagnosis at a more advanced stage in M (30). Accordingly, the poorer prognosis is not confirmed in the subgroup of patients affected by microcarcinomas (31). Less is known about treatment response by sex: in a retrospective study of 1,547 patients treated with surgery and/or radioiodine therapy, male sex was an independent risk factor for less than excellent responses to initial treatment (32). The role of sex and gender in the efficacy and safety of systemic treatment has not been studied (33). LEN is the first-line treatment in RR-DTC (20). Subgroup analysis of the SELECT trial confirmed that LEN was effective in increasing PFS in both sexes but efficacy endpoints and toxicities are not presented divided by sex (2). The study enrolled 138 patients (74 M and 64 F) well balanced for age, comorbidities, tumor characteristics, ECOG status and previous treatments. In our population, we found some gender differences in determinants of health: alcohol consumption and smoking habits were more frequent in men according to previous data (34, 35). Men were also more prone than women to perform physical activities, as already demonstrated in the younger non-oncological population (36). Compared to the SELECT trial, the median age of the patients included in our study was slightly higher, probably reflecting the real-world Italian setting. Only half of the patients received 24 mg of LEN as the starting dose despite the starting dose of 24 mg having been demonstrated as more effective than 18 mg (37). This could be caused by the higher age of our patients at the start of treatment, as supported by an inverse correlation between age and dose and the inclusion of 11% of patients with ECOG performance status of 2 or more.

Our study did not find sex differences in LEN toxicities, evaluated in terms of dose reduction, differently from what is reported in studies on other oncological treatments (7, 38). Almost all patients included in the study experienced at least one adverse event during LEN treatment. This is probably due to the long treatment duration but it is also in agreement with the literature data (16). The first side effect occurs, on average, after 5 weeks of treatment and the most frequent side effects were arterial hypertension and fatigue, as already reported in the phase III trial and real-world studies, both without providing subgroup analysis according to sex (2, 8, 9, 10, 11, 13, 16). Interestingly, the frequency of proteinuria was lower in our cohort compared with other studies (2, 16), possibly reflecting regional differences.

In our study, no sex difference was found in the number or the time of onset of side effects. Considering the type of adverse events, hematological side effects and mucositis were significantly more frequent in F. The higher risk of hematological side effects of TKIs in F has been observed in a study on 23,296 patients affected by many non-endocrine oncological diseases (7) but has also been confirmed, together with mucositis, in neuroendocrine neoplasms (38). The higher risk of hematological adverse events in F has already been reported in patients with colon-rectal cancers treated with irinotecan (39) or 5-fluorouracil-based adjuvant chemotherapy (40, 41, 42) and in patients affected by small-cell lung cancer treated with chemotherapy (43, 44).

The frequency of serious adverse events was high and occurred in 70% of patients, after an average of 9 weeks of treatment. The median dose at grade 3 adverse events was lower in elderly patients, who are likely more prone to develop toxicities, as confirmed by the higher frequency of grade 3 side effects and the lower dose of LEN in the elderly population compared to the young population (3). Considering the type of serious adverse events, gastrointestinal disorders were more frequent in M. These data differ from the aforementioned study by Unger, which reported a higher risk of gastrointestinal symptoms in F treated with TKIs (7). However, this study enrolled a heterogeneous population for primary site tumor, previous treatment and type of TKI. In our study, the most frequent serious adverse events were hypertension and weight loss, according to a meta-analysis of 561 patients treated with LEN for RR-DTC (44), while the incidence of fatigue was higher in our population.

Despite the high frequency of side effects, only 3.6% of patients discontinued treatment due to side effects, significantly lower than the 14.2% discontinuation rate in the SELECT trial (2). These data support the good safety profile of LEN, as also confirmed by a comparative study with sorafenib in RR-DTC (16).

Our study demonstrated the efficacy of LEN in both sexes, with an overall PFS (progression or death) of 29 months. Even if the comparison between different studies cannot be performed, PFS seems higher than in other studies (11, 45) including the pivotal study SELECT (2) (18.3 months), but similar to another real-world study (16). ORR was 36%, lower than what was reported in the SELECT trial (2) (64.8%) and in a meta-analysis of 13 studies on 554 patients (69%) (46). The presence of liver metastases was a negative prognostic factor for PFS and OS only in F. These data have never been reported, since no efficacy data stratified by sex are available in the literature. Although it is not possible to clarify the mechanism underlying these sex differences, the well-known sex differences in pharmacokinetics and pharmacodynamics (47) could play a role. In liver metastases, some isoforms of cytochrome P450 are lower than normal liver tissue and this is associated with shorter survival (48). The sex difference in cytochrome levels both in healthy and disease (49) and the higher proportion of fat mass in F (50) could cause a reduced clearance in F with liver metastases affecting survival.

The impact of only liver metastases, but not of bone and lung metastasis, on PFS and OS slightly differs from what is reported in the literature. A possible explanation is that liver metastases showed a lower median percentage change from baseline than lung and lymph nodes during LEN treatment (51) and enrolled patients with bone metastasis were treated with antiresorptive therapy such as zoledronic acid or denosumab. It is important to state that the primary objective of our study was the evaluation of sex differences in LEN adverse events assessed as dose reduction; therefore, the results on sex differences in PFS and OS should be carefully interpreted and need other studies for adequate confirmation.

The main limitation of the study is that it was retrospective. However, this allowed the enrollment of a high number of patients, with a broad regional representation (8 reference centers located in northern, central and southern Italy). It is not possible to exclude a sex bias also for investigators. Seventeen out of 20 authors of this study are females, while females are usually underrepresented in clinical oncological trials (52). Therefore, there is a potential influence of investigator gender both in clinical practice and in the way in which they collected data for experimental purposes (53), for example because females are usually more likely to listen to patients’ complaints. A strength of the study is the long duration of follow-up, which gave the possibility to evaluate OS and to better characterize the side effects and discontinuation rate of treatment with LEN. The main strength of the study is, for the first time, the choice of sex and gender difference assessment in both the safety and efficacy of LEN, with a rigorous study design according to SAGER guidelines (24), trying to fill a gap in the literature.

Conclusion

This is the first real-world study that, by stratifying results by sex, showed that LEN is effective in both males and females in radioiodine-refractory DTC. Despite the high frequency of side effects, LEN showed a good safety profile and a low discontinuation rate, with a sex difference in the frequency of some adverse events, with a high frequency of hematological disorders in males and mucositis in females. A better knowledge of sex and gender differences in the response to LEN, in the onset and characteristics of drug-induced toxicities is essential in order to improve its efficacy and safety, to ensure more appropriate and personalized therapy.

Declaration of interest

MA does consultancy and has received research grants from Bayer, Eisai and Eli-Lilly. LF is consultant for Eisai, Lilly, Ipsen. GPel does consultancy and has received research grants from EISAI, IPSEN, LILLY and IBSA. MGC does consultancy and has received research grants from EISAI-Lilly-IBSA-Merk. GS received honoraria from IBSA for participation as a speaker at scientific event. SZ does consultancy and has received research grants from Lilly, Ibsen, Eisai. GPul has been consultant for Bayer. CDur received honoraria from EISAI and Lilly for participation as a speaker at scientific and educational meetings. EA, AN, RL, MB, MM, SDL, IT, GG and CDal declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Funding

This work was financially supported through funding from the Institutional ‘Ricerca Corrente’ granted by the Italian Ministry of Health.

Author contribution statement

The authors confirm contribution to the paper as follows: M Appetecchia was responsible for the study conception and design. Data collection was carried out by C Giani, L Valerio, A Nervo, G Sapuppo, G Grani, C Dalmiglio, S De Leo, G Puliani, M Bianchini and S Zovato. The analysis and interpretation of results were conducted by I Terrenato, G Puliani, M Bianchini, M Normando and R Lauretta. The original draft of the manuscript was written by G Puliani and M Bianchini. Data validation and writing as well as review and editing were performed by S Zovato, L Fugazzola, M G Castagna, C Durante, G Pellegriti, E Arvat, R Elisei and M Appetecchia. M Appetecchia also provided supervision. All authors reviewed the results and approved the final version of the manuscript.