Comparison of secondary and primary thyroid cancer in adolescents and young adults


  • Melanie Goldfarb MD,

    Corresponding author
    1. University of Southern California Keck School of Medicine, Los Angeles, California
    • Corresponding author: Melanie Goldfarb, MD, Assistant Professor of Surgery, University of Southern California Keck School of Medicine, Division of Breast/Soft Tissue and Endocrine Surgery, Department of Surgery, 1510 San Pablo Street, Suite 412k, Los Angeles, CA 90033; Fax: (323) 865-3539;

    Search for more papers by this author
  • David R. Freyer DO, MS

    1. Children's Hospital Los Angeles, University of Southern California Keck School of Medicine, Los Angeles, California
    Search for more papers by this author



Thyroid cancer is one of the 5 most common malignancies in adolescent and young adult (AYA) patients (ages 15-39 years) and may develop de novo or in patients previously treated for cancer. This study compared the tumor characteristics, treatment, and overall survival (OS) of secondary malignant neoplasm (SMN) versus primary thyroid cancer in AYA patients.


All cases of AYA thyroid cancer contained in the 1998 to 2010 American College of Surgeons National Cancer Database were divided into 2 cohorts according to primary or secondary occurrence. Comparisons using appropriate statistical methods were performed.


Of 41,062 cases, 1349 (3.3%) had experienced a prior malignancy. Compared with cases of primary thyroid cancer, SMNs were more likely multifocal (odds ratio [OR] = 1.173, 95% confidence interval [CI] = 1.049-1.313) microcarcinomas < 1 cm (OR = 1.496, 95% CI = 1.327-1.687) with tall/columnar cells (OR = 2.187, 95% CI = 0.534-0.692), of white race (OR = 2.643, 95% CI = 1.310-5.331) and age 35-39 years (OR = 1.239, 95% CI = 1.093-1.404) and less likely female (OR = 0.608, 95% CI = 0.534-0.692), Hispanic (OR = 0.779, 95% CI = 0.642-0.946) age 15-19 years (OR = 0.624, 95% CI = 0.510-0.763) or 25-29 years (OR = 0.711, 95% CI = 0.604-0.837), or less likely > 4 cm in size (OR = 0.610, 95% CI = 0.493-0.758). There was a 6.63-fold (95% CI = 4.97-8.86, P < .001) relative risk of death for secondary versus primary thyroid cancers after adjusting for demographic, tumor, and thyroid treatment factors. Only Hispanic origin, tall/columnar cell histology, and distant metastases decreased OS for SMNs.


AYAs who develop thyroid cancer as a SMN have a significantly decreased OS compared to AYAs with primary thyroid cancer. Multiple demographic and tumor differences exist between these 2 cohorts. Whether the outcome disparity results from previous cancer treatment or differences in biology, environment, or access to care are areas needing further investigation. Cancer 2014;120:1155–1161. © 2014 American Cancer Society.


Thyroid cancer is one of the 5 most common malignancies in adolescent and young adults (AYAs, defined as those aged 15-39 years) and may develop de novo or in individuals previously treated for cancer. The most recent SEER (Surveillance, Epidemiology and End Results) incidence rates show that for female AYAs, thyroid cancer is the most common cancer in age groups 15 to 19, 20 to 24, and 25 to 29 years and the second most common cancer for ages 30 to 34 and 35 to 39 years.[1] Overall, it is the fourth most common cancer for individuals aged 15 to 19 years, but the second most common for those aged 20 to 39 years. Although the majority are primary thyroid cancers, a subset are secondary malignant neoplasms (SMNs) arising in survivors of pediatric or other AYA malignancies including acute lymphocytic leukemia, central nervous system tumors, or bone and soft tissue sarcomas. The greatest risk is for patients diagnosed at younger ages (≤ 17 years for acute lymphocytic leukemia and central nervous system tumors; ≤ 18 years for bone and soft tissue sarcomas).[2-4] These patients may have experienced toxic and environmental exposures such as chemotherapy and/or radiation therapy, which have the potential to cause secondary cancers during their young adult life.[5, 6] Influences such as family history, lifestyle behaviors, and comorbid health conditions may also play a role in secondary cancer development for AYAs.[7, 8] However, little is known about how these associations apply to thyroid cancer. Therefore, the purpose of this study is to determine whether demographic, clinical, pathologic, or survival differences exist between primary and secondary thyroid malignancies in AYA patients.


Data Source

The National Cancer Database (NCDB) is a joint project of the American College of Surgeons (ACoS) Commission on Cancer and the American Cancer Society. The NCDB contains data elements on patient demographics, insurance status, tumor characteristics, first course of treatment, ZIP code–level socioeconomic factors, and facility-level characteristics. A recent analysis showed that 76.1% of all thyroid cancers in the United States are included in this database and 71.1% to 77.4% of AYA cancers.[9] Data reporting to NCDB is highly standardized and similar to other federal cancer registry data systems. After a patient was diagnosed and treated at an ACoS-accredited cancer program, the hospital registrar was responsible for documenting care, even when the patient was transferred to another facility.[10] The data are then coded and reported according to nationally established protocols coordinated by the North American Association of Central Cancer Registries.[11] Because all patient information is deidentified, this study was determined by our institutional review board to be exempt from approval.

Study Cohort and Measures

All 41,062 cases of AYA thyroid cancer contained in the 2004-2010 NCDB were included in this study. Patients were divided into 2 cohorts according to whether their thyroid cancer was a primary or SMN. Demographic variables included age, sex, race, ethnicity, insurance and socioeconomic status. Age categories within the AYA population were divided into 5-year intervals as follows: ages 15 to 19, 20 to 24, 25 to 29, 30 to 34, and 35 to 39 years. Insurance status was classified as “uninsured” or “any insurance,” and low socioeconomic status (SES) as < $46,000 annual income, as defined by the NCDB. Patient race was categorized by the NCDB as white, black, Asian/Pacific Islanders, Native American, or other, and ethnicity as Hispanic or non-Hispanic. Race and ethnicity were included in the analysis because they have been shown to influence outcome following treatment/response to treatment.[12] Tumor characteristics included histology (differentiated thyroid cancer [DTC] including all variants versus non-DTC), grade (moderate-well differentiated versus poor or undifferentiated that is coded as a separate variable than histological subtype), tumor focality (solitary or multiple nodules), tumor size (< 1 cm, > 4 cm), lymph node status (unknown or negative lymph nodes versus documented positive nodes), and distant metastases at presentation (present or absent; unknown were excluded). Treatment variables included extent of surgery (total thyroidectomy or less than total thyroidectomy), any lymph nodes examined, postoperative radioiodine (RAI) or radiation (XRT), and less than 24-hour hospital stay. Last date of contact was used for the amount of follow-up time; those with “0” months or missing follow-up data were excluded from the survival analyses.

Statistical Analysis

Nominal categorical variables were compared using Fisher exact test and ordered categorical variables were analyzed using the Mantel-Haenszel chi-square test to analyze the demographic, pathologic, and treatment characteristics of the study cohort. Stepwise binary logistic regression was employed to identify independent factors associated with primary or secondary thyroid cancer. The model goodness of fit was assessed by evaluating model discrimination (c-statistic) and calibration (Hosmer-Lemeshow chi-square). Odds ratios and 95% confidence intervals (CIs) were calculated to evaluate the strength of association between each variable and the occurrence of a secondary thyroid malignancy. Kaplan-Meier analysis with the log-rank test was used to determine if any demographic or clinical variable differences significantly impacted overall survival (OS). Analyses were also performed after stratification by primary or secondary malignancy status. Cox proportional hazards regression modeling with forward regression was performed to identify independent factors associated with OS. All variables that were significant (P < .1) on univariate Kaplan-Meier analysis in addition to tumor sequence were entered into the regression model, and only those with a P < .05 were retained in the final model. Hazard ratios and 95% CIs were calculated to evaluate the strength of association between each variable and survival.

Data analyses were performed with the Statistical Package for the Social Sciences (SPSS) software, version 20.0 (SPSS, Chicago, IL) and SAS version 9.3 (SAS Institute Inc, Cary, NC); all tests were 2-sided, and a P value of < .05 was considered to be statistically significant.


Of 41,062 AYA thyroid cancer cases, 1349 (3.3%) had experienced a prior malignancy. Within this cohort of SMNs, 23.7% were male, 90.5% were white, 8.5% were Hispanic, and 13.8% between ages 15 and 25 years. Most cases (83.2%) had some form of insurance, although 55% were considered of low SES. Compared with cases of primary thyroid cancer, those with secondary cancers had a higher proportion of whites (P < .001), males (P < .001), non-Hispanics (P = .001), and significantly different age (P < .001) and race distributions (P < .001). Secondary cancers were more common in 15- to 19-year-olds and 35- to 39-year-olds and less common in 20- to 24-year-olds and 25- to 29-year-olds as well as in Asians and those of other race. Pathologic tumor (P < .001) and nodal (P = .048) staging was different between the 2 groups: secondary cancers were smaller (both microcarcinomas < 1 cm [P < .001] and < 4 cm [P < .001]) but more likely multifocal (P = .005) and with tall/columnar cell histology if a DTC. There was no difference in the frequency of positive lymph nodes in patients who had nodes examined, the proportion of DTC histology, or moderate to well-differentiated grade (P = not significant [NS]). In addition, there was no difference in rates of distant metastases at presentation between groups (P = NS). There was also no significant difference in extent of surgery or use of external beam radiation between the 2 cohorts. However, patients with secondary cancers waited a mean of 2.7 days longer (P = .009) for surgery and were less likely to receive RAI therapy (P = .0005) (Table 1).

Table 1. Patient and Disease Characteristics by Type of Thyroid Cancer Occurrence
n = 41,062 (unless otherwise noted)Primary Thyroid Cancer (%)Secondary Thyroid Cancer (%)P
Male sex16.423.7<.001
Age category, y  <.001
Ethnicity  <.001
American Indian0.30.1.061
Hispanic origin11.48.5.001
Low socioeconomic status56.055.0.45
Tumor Characteristics
Papillary/follicular histology84.786.3.12
Tall/columnar cell variant1.12.1.001
Well-differentiated (n = 6667)96.396.5.92
Size <1 cm27.338.3<.001
Size >4 cm12.16.9<.001
Positive lymph nodes27.926.1.16
Distant metastases (n = 40257)
Surgical treatment   
Any nodes examined58.456.2.11
Total thyroidectomy86.186.8.26
Postoperative therapy   
Beam radiation0.30.2.51
<24-hour stay78.877.0.11

Stepwise multivariate regression demonstrated that SMNs occurred more commonly in whites and in individuals aged 35 to 39 years. SMNs were more likely microcarcinomas < 1 cm and multifocal with tall/columnar cell histology (model c-statistic 0.631, 95% CI = 0.617-0.646; Hosmer-Lemeshow chi-square = 0.702). Patients with secondary thyroid cancers were less likely female, Hispanic, ages 15 to 19 or 24 to 29 years at the time of this diagnosis, and less likely to have a tumor size > 4 cm (Table 2).

Table 2. Multivariate Predictors of Secondary Thyroid Cancers
Patient (N = 41,062) CharacteristicsOdds Ratio95% Confidence IntervalP
Female sex0.6080.5340.692<.001
Tall/columnar cell histology2.1871.4893.210<.001
Age 15-19 y0.6240.5100.763<.001
Age 25-29 y0.7110.6040.837<.001
Age 35-39 y1.2391.0931.404.001
White race2.6431.3105.331.007
Tumor size <1 cm1.4961.3271.687<.001
Tumor size >4 cm0.6100.4920.758<.001
Hispanic origin0.7790.6420.946.012

Mean follow-up after thyroid cancer diagnosis was 35.5 ± 24.0 months (range, 1-96.1 months) and was not different between those with primary (35.5 ± 24.1 months) and secondary (34.9 ± 23.1 months) DTC. Multiple demographic, pathologic, and clinical treatment factors affected OS on univariate Kaplan-Meier analyses (Table 3). However, after stratification by secondary malignancy status, Hispanic origin, other race, DTC histology, tall/columnar cell histology, thyroid hormone replacement, any lymph nodes examined, and distant metastases at presentation negatively affected OS on univariate analysis. (Table 4).

Table 3. Univariate Predictors of Overall Survival
 NMean Survival (mo)P (Log-Rank)
Tumor order  .569
Primary thyroid cancer3955235.689 
Secondary thyroid cancer136735.919 
Sex  .17
Race  <.001
American Indian13335.878.426
Socioeconomic status  .002
Insurance status  <.001
Hispanic origin  <.001
Age category, y  <.001
Metropolitan location  .005
Tumor Characteristics
Tumor differentiation  .507
Poorly differentiated24435.324 
Differentiated thyroid carcinoma histology  <.001
Tall/columnar cell histology   
Tumor focality  .059
Tumor Characteristics
Tumor size >4 cm   
Microcarcinoma (<1 cm)  <.001
Documented pathologic lymph node involvement  <.001
Distant Metastases  .01
Treatment Factors
Extent of surgery  <.001
Total thyroidectomy (TTx)35,37735.517 
Less than TTx568536.833 
Any lymph nodes examined  <.001
Radioiodine therapy  <.001
External beam irradiation  <.001
Thyroid hormone replacement  <.001
Table 4. Univariate Predictors of Survival for Patients With Secondary Thyroid Cancers
PredictoraNMean Survival (mo)P (Log-Rank)
  1. a

    Only significant predictors (P < .1) are displayed.

Other race  .034
Hispanic origin  .045
Distant metastases at presentation  .01
Tall/columnar cell histology  .037
Differentiated thyroid carcinoma  .004
Thyroid hormone replacement  .017
Any lymph nodes examined  <.001
Tumor size >4 cm  .077
Positive lymph nodes  .085

Multivariate Cox regression showed that for all AYA patients with thyroid cancer, female sex and RAI therapy improved OS whereas secondary malignancy status, Hispanic ethnicity, low SES, age 35 to 39 years, having lymph nodes examined or positive lymph nodes, and distant metastases at presentation decreased OS (Table 5). The remaining variables (insurance status, metropolitan treatment location, DTC, tumor focality and size < 1 cm, total thyroidectomy, XRT, thyroid hormone replacement, white and Asian race, and ages 15-19, 20-24, and 25-29 years) did not reach significance in the final model. After adjusting for demographic, tumor, and thyroid treatment factors, there was a still a 6.63-fold (95% CI = 4.97-8.86; P < .001) relative risk of death for secondary versus primary thyroid cancers (Fig. 1). In addition, for those with a secondary thyroid malignancy, only Hispanic ethnicity, tall/columnar cell histology, and distant metastases decreased OS (Table 5).

Table 5. Multivariate Cox Regression Predictors of Decreased Overall Survival
Patient CharacteristicsHazard RatioP95.0% Confidence Interval
  1. Abbreviation: AYA, adolescent and young adult.

All AYA Patients With Thyroid Cancer (N = 39,139)
Hispanic origin2.214<.0011.6802.918
Lymph nodes examined1.376.0251.1031.933
Positive lymph nodes1.382.0271.0381.8240
Low socioeconomic status1.921<.0011.5042.454
Radioiodine therapy0.502<.0010.4000.631
Second malignancy6.438<.0014.8278.587
Distant metastases at presentation9.820<.0015.32218.120
Female sex0.591<.0010.4590.762
Age 35-39 y1.489.0011.1851.871
Only Patients With Secondary Thyroid Cancer (N = 1326)
Hispanic origin4.230<.0012.2667.898
Distant metastases at presentation13.032<.0014.01042.350
Tall/columnar cell histology5.295.0011.89814.773
Figure 1.

Product-limit overall survival estimates are shown for primary and secondary adolescent and young adult thyroid cancers (n = 39,693). Abbreviations: Dx, diagnosis; HR, hazard ratio.


This study indicates that secondary malignancy status is an independent risk factor for decreased OS among AYAs diagnosed with thyroid cancer. Patients for whom thyroid cancer is a SMN have a 6.63-fold risk of death independent of demographic, tumor, and treatment factors compared to patients with primary thyroid cancer over a short follow-up period. This is an important new observation for the 3% to 4% of this population who are already survivors of pediatric or other AYA cancers. These numbers are not insignificant considering that thyroid cancer is one of the most common AYA cancers, especially among females, and patients need to be counseled appropriately. In addition, differences in thyroid cancer survival usually require 15 to 20 years of follow-up, rendering the results of this study quite significant. Future research should explore possible disease-specific, biological, environmental, or access-to-care explanations for these findings. In addition, because the difference in OS among patients with primary and secondary malignancy over a short follow-up period persisted even after controlling for various clinical and thyroid cancer treatment factors, the overall treatment strategy for prior cancer survivors with a secondary DTC should be a focus for future clinical investigation.

Only one study has looked at differential gene expression patterns in AYAs compared to older individuals.[13] The authors hypothesized that differences in the extent of disease at presentation and survival of patients with papillary thyroid carcinoma that existed between AYAs and older patients have a biological/genetic basis. However, they were unable to find a difference in the frequency or type of somatic mutations between the 2 groups, although 6 genes (extracellular matrix protein 1 [ECM1], v-erb-2 erythroblastic leukemia viral oncogene homolog 2 [ERBB2], urinary plasminogen activator [UPA], 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase 2 [PFKFB2], meis homeobox 2 [MEIS2], and carbonic anhydrase II [CA2]) had significant differential expression between the AYA and older patients. They did not comment on SMNs in either group of patients.

Primary thyroid cancers occurred more often in AYAs in their 20s compared to those who were 15 to 19 years old or over the age of 30 years, where secondary thyroid cancers were more common. This may be a consequence of a long latency in developing clinically apparent thyroid cancer after radiation or chemotherapy, which may be up to 30 years after initial pediatric cancer treatment. Alternatively, older patients have had more time to develop an assortment of secondary thyroid conditions such as multinodular goiter or hyperthyroidism and finally undergo thyroidectomy as treatment for these conditions, only to discover incidental, secondary thyroid cancer in a final pathology specimen. Not surprisingly, age category within AYAs did not impact OS on multivariate Cox regression, inasmuch as the prognosis is excellent for patients with thyroid cancer under the age of 45 years.[5, 14, 15]

In AYA patients with thyroid cancer, SMNs were more likely microcarcinomas < 1 cm and less likely to have a tumor > 4 cm. This could be a consequence of increased surveillance and thyroid screening in previous cancer survivors. In North America, monitoring of childhood cancer survivors is largely guided by the Children's Oncology Group (COG) Long-Term Follow-Up Guidelines for Survivors of Childhood, Adolescent and Young Adult Cancers.[16, 17] These risk-adapted guidelines are developed on the basis of published evidence where available and supported by expert consensus opinion. For childhood cancer survivors who received irradiation to or in proximity with the thyroid gland, current surveillance recommendations consist of an annual physical examination; ultrasound and fine-needle aspiration (FNA) or biopsy are limited to patients who have palpable nodules. Because microcarcinomas would generally not be discovered by palpation, the most plausible explanations are that clinicians are screening more aggressively with ultrasound than current COG guidelines recommend, that small nodules are detected as part of routine surveillance for the patient's original primary tumor, or that these microcarcinomas are only discovered incidentally when the thyroid is removed for another reason, such as hyperthyroidism or multinodular goiter. In regard to screening practices for secondary thyroid cancer, it is of interest that in the current COG guidelines (version 3.0) routine thyroid ultrasound is no longer recommended as it was in past versions because of the high frequency of FNA, biopsy, and thyroidectomy triggered by nonpalpable nodules that later proved to be histologically benign. The increased prevalence and negative impact of tall/columnar cell histology in SMNs is an interesting finding that could be a consequence of prior cancer treatments causing dedifferentiation of latent DTCs or an inherent biological/genetic tendency to develop more aggressive tumors as a SMN. Prior studies in Chernobyl (Ukraine), Hodgkin's disease, and Childhood Cancer Surveillance Study cohorts that have some SMN thyroid cancers after radiation exposure did not comment on tall/columnar cell variants.

Patients with secondary thyroid malignancies were also more likely to be male, white, and non-Hispanic. However, only Hispanic ethnicity, tall/columnar cell histology, and distant metastases at presentation decreased OS for these patients. Sex, ethnicity, and SES also affected OS for the entire cohort, similar to previous studies in other thyroid cancer cohorts.[14, 15, 18] In addition, age 35 to 39 years and certain tumor characteristics decreased survival in the larger group. Whether biological or environmental explanations for these characteristics make these patients more vulnerable to a second malignancy or worse overall outcomes is an area for further study.

A recent study developed and validated a risk model, based on self-reported data, for developing secondary thyroid cancer in survivors who had been diagnosed with cancer at less than 18 years of age.[19] In that population, females, those with prior thyroid nodules, primary cancer before age 15, treatment with an alkylating chemotherapy and/or radiation, and birth year after 1970 all contributed to a patient's risk for developing secondary thyroid cancer. Interestingly, type of primary cancer was not included in the final model. History of thyroid nodules conferred the greatest risk for subsequent thyroid cancer development, which may have reflected more frequent referral for surgery and detection of incidental microcarcinomas. Unfortunately, the report did not provide information on the tumor size or clinical significance of the diagnosed thyroid cancers. Replicating these findings in AYA patients who receive radiation as part of an initial cancer treatment would be a valuable addition toward providing screening guidelines for the AYA cancer survivor population and might provide insight into the increased prevalence of a thyroid SMN in 35- to 39-year-olds.

Although the NCDB captures the largest number of cancer cases in the United States, long-term survival data are presently limited due to the infancy of the database. Therefore, this study's reported survival outcomes will need to be verified with another large database that has longer follow-up data, or at a later date when the NCDB has more follow-up information. Disease-free survival is also not part of the NCDB and would be another important clinical outcome variable to measure and may be not as significant as the difference in OS, because thyroid cancer has an overall excellent prognosis. Moreover, for patients with secondary thyroid cancers, no information is available on the primary tumor type or prior cancer treatment exposures, which and could influence both thyroid cancer characteristics and survival. In addition, causes of death are not captured in the NCDB. Although the present study appeared to have an appropriate distribution of ethnicities (11% of the patients were Hispanic), the NCDB has been shown to only capture 51.1% of Hispanic cancer cases, so that data may be skewed.[9] Also, other postoperative outcomes specific to thyroid surgery that influence disease morbidity, such as hypoparathyroidism or recurrent laryngeal nerve injury, are not captured in any database but would be important considerations in evaluating the role of thyroid procedures in managing AYAs with secondary thyroid nodules.


AYAs that develop thyroid cancer as a SMN have a significantly decreased OS compared to AYAs with primary thyroid cancer. Secondary malignancies were more likely microcarcinomas < 1 cm, multifocal, with tall/columnar histology and occurred more often in males, non-Hispanics, those of nonwhite race, and AYAs in their 20s. In addition, OS for secondary thyroid cancers was impacted by Hispanic origin, tall/columnar cell histology and the presence of distant metastases at presentation. Whether this is a direct consequence of previous cancer treatment, biological or environmental factors, or disparity in access to care is an area that needs further investigation.


No specific funding was disclosed.


The authors made no disclosure.