Keywords thyroid cancer - secondary primary malignancy - lymphoma - leukemia
Introduction
According to the International Agency for Research on Cancer’s GLOBOCAN 2020 Cancer
Incidence and Mortality database, thyroid cancer is the ninth most common cancer
worldwide, with 43 646 new cases in 2020 [11 ]. The number of new cases in the U.S. increased steadily from 1978 to
2009 and has now reached a relatively stable plateau. At the same time, the overall
very good 5-year survival rate has improved from 93.22% to 98.59% [22 ]. Although thyroid cancer can occur in
people of any gender, the incidence is three times higher in women than in men [11 ]
[22 ]. Notably, this gender distribution shows no regional differences [11 ]. Mortality is also almost twice as high
in women as in men [22 ]. The peak incidence
of thyroid cancer is between the ages of 50 and 64 [22 ]. Interestingly, thyroid cancer is the
second most common malignancy in female adolescents and adults between the ages of
15 and 39, and the fourth most common in men [22 ]. Histopathologically, thyroid cancer is broadly classified into
differentiated thyroid cancer (DTC), undifferentiated thyroid cancer (UTC), and
medullary thyroid cancer (MTC) [33 ]. DTCs
include papillary thyroid carcinoma, which is the most common malignant thyroid
tumor (approximately 84%), follicular thyroid carcinoma (FTC, approximately 4%), and
oncocytic thyroid carcinoma (approximately 2%). In contrast, biologically aggressive
UTCs such as anaplastic thyroid carcinoma (ATC, about 2%) and medullary thyroid
carcinoma originating from parafollicular C-cells (about 4%) are less common. While
surgical resection with lymphadenectomy is the mainstay of treatment for all
localized thyroid cancers, the therapeutic options for ATC and MTC that have
metastasized outside the neck or are unresectable are very limited and include
radiochemotherapy for ATC or combined BRAF/MEK inhibitor therapy in the presence of
a BRAFV600E mutation and treatment with tyrosine kinase inhibitors for MTC [33 ]
[44 ].
For a long time, complete thyroidectomy followed by ablation of all remaining thyroid
tissue with radioiodine (RAI) therapy was considered the best treatment for DTC
[55 ]
[66 ]. However, this paradigm has changed in
recent years in some countries. According to the American Thyroid Association (ATA),
DTCs are classified into low, intermediate, and high risk tumors based on specific
characteristics and the associated risk of recurrence [77 ]. Although adjuvant RAI improves overall
survival and disease-free survival in advanced DTC, most studies show little or no
benefit of RAI in low and intermediate risk tumors [88 ], where 5-year recurrence-free survival
without RAI is already 97% [99 ]. While
approximately 54% of all patients with DTC are in a low-risk situation for which
routine RAI therapy is not recommended according to ATA recommendations, it is
suggested under certain conditions for intermediate-risk situations (< 38%) and
even recommended for high-risk patients (< 8%). In this context, the goal of RAI
therapy is the ablation of residual thyroid tissue, adjuvant therapy, or the
treatment of manifest tumor disease. However, since the introduction of RAI therapy,
concerns have been raised about a possible carcinogenic effect. A meta-analysis of
13 studies [1010 ] found an increased
incidence of secondary primary malignancies (SPMs) in patients with DTC, which may
be related to disease-specific therapy or genetic predisposition. In addition, the
results of a previous meta-analysis that included 2 multicenter studies [1111 ] suggest that the occurrence of SPMs in
DTC may be associated with the use of RAI. As thyroid cancer is one of the most
common tumors, especially in adolescents and adults between 15 and 39 years of age,
these patients could be particularly affected by the consequences of a SPM.
Given the lack of population-based studies on this topic, the aim of our
retrospective cohort study was to investigate the association between thyroid cancer
and the incidence of subsequent lymphoma and leukemia.
Materials and Methods
Database
The study is based on data from the Disease Analyzer database (IQVIA), which
contains drug prescriptions, diagnoses, and basic medical and demographic data
obtained directly and anonymously from the computer systems used in the
practices of general practitioners and specialists [1212 ]. The database includes
approximately 1300 general practices in Germany. The panel of practices included
in the Disease Analyzer database has previously been shown to be representative
of general and specialist practices in Germany [1212 ]. Finally, this database has been
used in previous studies focusing on cancer [1313 ]
[1414 ].
Study population
This retrospective cohort study included adults (≥18 years) with a first
documented diagnosis of thyroid cancer (ICD-10: C73) in 1284 general practices
in Germany between January 2005 and December 2021 (index date; [Fig. 1Fig. 1 ]). A further inclusion
criterion was an observation period of at least 12 months prior to the index
date. Patients with other cancer diagnoses (ICD-10: C00-C97 excl. C73) before or
on the index date were excluded.
Fig. 1
Fig. 1 Selection of study patients.
Thyroid cancer patients were matched (1:5) to non-thyroid cancer patients using
nearest neighbor propensity scores based on age, sex, index year, and annual
visit frequency during follow-up. Because thyroid cancer patients have much
higher consultation frequency, and higher consultation frequency may increase
the likelihood of documentation of other diagnoses, we included consultation
frequency per year in the matching.
Non-thyroid cancer individuals were included only if they had an observation
period of at least 12 months prior to the index date and no cancer diagnoses in
their medical history before or on the index date. For non-thyroid cancer
patients, the index date was the date of a randomly selected visit between
January 2000 and December 2021 ([Fig. 1Fig.
1 ]).
Study outcomes
The study outcome was the incidence of lymphoma and leukemia within 10 years of
the index date as a function of thyroid cancer. Lymphomas included Hodgkin
lymphoma (ICD-10 C81), follicular lymphoma (ICD-10 C82), non-follicular lymphoma
(ICD-10 C83), mature T/NK cell lymphoma (ICD-10 C84), other specified and
unspecified types of non-Hodgkin’s lymphoma (ICD-10 C85), other specified types
of T/NK-cell lymphoma (ICD-10 C86), malignant immunoproliferative disorders, and
certain other B-cell lymphomas (ICD-10 C88). Leukemias included lymphoid
leukemia (ICD-10 C91), myeloid leukemia (ICD-10 C92), monocytic leukemia (ICD-10
C93), other leukemia of specified cell type (ICD-10 C94), leukemia of
unspecified cell type (ICD-10 C95).
Statistical analyses
Differences in sample characteristics between those with and without thyroid
cancer were tested using the Wilcoxon signed-rank test for continuous variables,
the McNemar test for categorical variables with two categories, and the
Stuart–Maxwell test for categorical variables with more than two categories.
Univariate Cox regression models were performed to examine the association
between thyroid cancer and the incidence of subsequent lymphoma and leukemia
diagnoses. These models were performed separately for four age groups, women,
and men. To address the issue of multiple comparisons, p-values <0.01 were
considered statistically significant. Analyses were performed with SAS version
9.4 (SAS Institute, Cary, USA).
Results
Baseline characteristics of the study cohort
The present study included 4232 individuals with thyroid cancer (mean age: 54.2
years; 73.6% female) and 21 160 individuals without thyroid cancer (mean age:
54.2 years; 72.6% female). Baseline characteristics of the study patients are
shown in [Table 1Table 1 ]. On average,
patients visited their primary care physician 7.9 times per year during the
follow-up period.
Table 1
Table 1 Basic characteristics of the study sample (after
1:5 propensity score matching).
Variables
Proportion affected among individuals with thyroid cancer (%)
N=4232
Proportion affected among individuals without thyroid cancer
(%) N=21160
p-Value
Age (Mean, SD)
54.2 (15.4)
54.2 (15.4)
1.000
Age≤50
840 (19.9)
4200 (19.9)
1.000
Age 51–60
877 (20.7)
4385 (20.7)
Age 61–70
1,034 (24.4)
5170 (24.4)
Age>70
1481 (35.0)
7405 (35.0)
Women
3113 (73.6)
15 359 (72.6)
0.194
Men
1119 (26.4)
5801 (27.4)
Yearly consultation frequency during the follow-up period
7.9 (4.1)
7.9 (4.1)
1.000
Year of index date
2005–2008
489 (11.6)
2542 (12.0)
0.206
2009–2012
806 (19.1)
3834 (18.1)
2013–2016
1141 (27.0)
5529 (26.1)
2017–2021
1796 (42.4)
9255 (43.7)
Proportions of patients in% given, unless otherwise indicated. SD:
standard deviation.
Association between thyroid cancer and lymphoma incidence
[Fig. 2Fig. 2 ] shows the Kaplan–Meier curves
for the time to first lymphoma diagnosis with an incidence of 1.29 cases per
1000 person-years in patients with thyroid cancer and 0.39 cases per 1000
person-years in patients without thyroid cancer. The most common lymphoma
diagnosis was other specified and unspecified non-Hodgkin’s lymphoma (ICD-10
C85) (57.1% of lymphoma cases in the thyroid cancer cohort and 62.9% in the
non-thyroid cancer cohort). In Cox regression analyses, thyroid cancer was
strongly and significantly associated with lymphoma incidence [Hazard ratio
(HR): 3.35; 95% CI: 2.04–5.52]. This association was stronger in men (HR: 5.37;
95% CI: 2.22–12.98) than in women (HR: 2.67; 95% CI: 1.45–4.94). In the
age-stratified analysis, the strongest association was observed in the age group
61–70 years but did not reach the predefined significance level of p<0.01
([Table 2Table 2 ]).
Fig. 2
Fig. 2 Kaplan–Meier curves for time to lymphoma diagnosis in
individuals with and without thyroid cancer.
Table 2
Table 2 Association between thyroid cancer and the
incidence of lymphomas and leukemias in individuals followed in
general practices in Germany (Cox regression models).
Lymphomas
Leukemias
Patient group
HR (95% CI)
p-Value
HR (95% CI)
p-Value
Total
3.35 (2.04–5.52)
<0.001
1.79 (0.91–3.53)
0.087
Age≤50
1.11 (0.11–11.03)
0.928
4.19 (0.59–29.87)
0.153
Age 51–60
4.33 (1.25–15.00)
0.021
4.47 (0.28–71.63)
0.290
Age 61–70
5.45 (1.46–20.34)
0.012
1.35 (0.36–5.03)
0.653
Age>70
3.13 (1.67–5.88)
<0.001
1.61 (0.63–4.11)
0.322
Women
2.67 (1.45–4.94)
0.002
1.33 (0.53–3.33)
0.546
Men
5.37 (2.22–12.98)
<0.001
2.81 (1.00–7.93)
0.051
Association between thyroid cancer and leukemia incidence
[Fig. 3Fig. 3 ] shows the Kaplan–Meier curves
for the time to first leukemia diagnosis with an incidence of 0.55 cases per
1000 person-years in patients with thyroid cancer and 0.31 cases per 1000
person-years in patients without thyroid cancer. Overall, the number of patients
with a leukemia diagnosis was very small (12 cases in the thyroid cancer cohort
and 28 cases in the non-thyroid cancer cohort), with lymphoid leukemia being the
most common leukemia diagnosis (58.3% of leukemia cases in the thyroid cancer
cohort and 35.7% in the non-thyroid cancer cohort). In Cox regression analyses,
thyroid cancer was not significantly associated with the incidence of leukemia
(HR: 1.79; 95% CI: 0.91–3.53). The nonsignificant association was strong in the
age groups≤50 years (HR: 4.19; 95% CI: 0.59–29.87) and 51–60 years (HR: 4.47;
95% CI: 0.28–71.63), but due to the small number of patients with leukemia, the
associations were not significant ([Table 2Table
2 ]).
Fig. 3
Fig. 3 Kaplan–Meier curves for time to leukemia diagnosis in
individuals with and without thyroid cancer.
Discussion
Our results show a significant association between thyroid cancer and the incidence
of lymphoma, which was particularly high among men. However, this association was
also observed in women and in different age groups, although with different
statistical significance. The incidence of lymphoma was significantly higher in
thyroid cancer patients than in controls, suggesting a possible link between thyroid
cancer and the subsequent development of lymphoma. This finding also underscores the
importance of careful monitoring and tailored surveillance strategies for thyroid
cancer patients, particularly for the early detection of lymphoma.
Conversely, the study found no significant association between thyroid cancer and
the
incidence of leukemia, although there was a trend toward an increased incidence,
especially in younger age groups. However, it should be noted that the small number
of leukemia cases may have reduced the statistical power to detect significant
associations. Nevertheless, the observed trend suggests the need for further
investigation of the possible association between thyroid cancer and leukemia,
especially in younger patient populations.
The incidence of thyroid cancer has increased worldwide in recent decades,
disproportionately affecting women and East Asia more than other regions [1515 ]. Numerous studies have shown that this
global increase in thyroid cancer incidence is due in large part to the improved
detection of small, low-risk papillary thyroid cancers made possible by the
widespread use of thyroid ultrasound [1515 ]
[1616 ].
As a result of diagnostic and therapeutic advances in the treatment of thyroid
cancer, including RAI therapy for DTC, overall survival from thyroid cancer has
improved significantly over the past two decades, with the 5-year survival rate
increasing from 93.22% in 1978 to 98.59% in 2009 [22 ]. In parallel with the growing number of
cancer survivors [1717 ] in recent decades,
the incidence of SPMs has also been rising [1818 ]. In a large population-based study, Donin and colleagues demonstrated
that one in twelve survivors of a common cancer developed a SPM, with lung cancer
being the most commonly diagnosed SPM [1919 ]. Furthermore, in this group, more than half (55%) died from their SPM,
exceeding the proportion of patients with only one cancer who died from it [1919 ].
Thus, the burden of SPMs in a growing and aging population of primary cancer
survivors has increased significantly in recent decades. Cancer survivors may be
predisposed to the development of SPMs by a variety of factors, including cancer
predisposition syndromes or specific tumor characteristics, environmental exposures,
and late effects of therapies [2020 ].
Knowledge of the predisposition to develop a SPM and its type is relevant for
tailoring tumor follow-up. Thus, the meta-analysis of 13 studies published by
Subramanian et al. showed that the incidence of SPMs in the form of non-Hodgkin’s
lymphoma (NHL), multiple myeloma (MM), and leukemia was significantly increased in
thyroid cancer survivors [1010 ]. However,
the results of individual studies are very heterogeneous. While some studies found
an increased risk of Hodgkin’s lymphoma (HL), NHL, and leukemia after thyroid cancer
[2121 ], others reported either an
association with NHL and leukemia [2222 ] or
even an association with subsequent leukemia but not lymphoma [2323 ]. Some of these studies also examined
whether there was an association between the use of RAI and the incidence of
lymphatic and hematopoietic malignancies [2121 ]
[2323 ]. Brown et al. showed an
increased risk of MM and HL after RAI, while patients who did not receive RAI were
more likely to develop leukemia [2121 ].
Rubino and colleagues, on the other hand, described an increased relative risk of
leukemia after RAI [2323 ]. A study published
in 2017 demonstrated that in the group of patients with DTC who frequently developed
ALL or MM after surgery alone, the risk of developing AML, CLL or CML was increased
by RAI [2424 ]. In this context, there
appears to be a dose dependence, as a recent nationwide population-based study from
South Korea showed that RAI above 100 mCi was strongly associated with the
development of leukemia, whereas this was not the case for lower RAI doses [2525 ]. Given the favorable survival rates of
patients with DTC and the concerns about the potentially adverse effects of RAI,
physicians should weigh the benefits and risks of RAI before choosing this
therapeutic strategy. A useful tool for determining the risk of recurrence and
therefore the benefit of adjuvant RAI is the ATA recommended risk classification of
DTCs [77 ].
Interestingly, however, only the study by Brown et al. examined the association
between thyroid cancer and lymphatic or hematopoietic tumors separately for men and
women and in different age groups [2121 ]. In
contrast to this study, we were able to demonstrate a stronger association in men.
However, the observation that the incidence of developing SPM was increased when
thyroid cancer was diagnosed in middle age was comparable.
In addition to RAI, environmental factors such as low-dose radiation used in
diagnostic procedures have also been associated with an increased risk of thyroid
cancer as well as lymphatic and hematopoietic tumors [2626 ]. Although DTCs certainly account for
the highest proportion of thyroid cancers in our cohort, but are known to have a low
mutational burden [2727 ], genetic factors
may also explain the association we observed.
A major strength of our study is the large sample size. In fact, our study is one
of
the largest population-based studies on this topic and the only one to date that
specifically examined the outpatient setting. In addition, our study excluded
individuals with other tumor diagnoses on the index day. Moreover, the cohort was
matched to the control cohort with respect to age, sex, index year, and even the
average number of consultations per year. In particular, including of consultation
frequency in matching reduces the influence of confounding factors such as
utilization patterns.
But we must also note that our study has several limitations, especially in the study
design. As diagnoses are based on ICD codes, misclassification cannot be excluded.
Unfortunately, we were also unable to differentiate between the common
histopathologic subtypes of thyroid cancer. Furthermore, as the database only
contains data from outpatients, and no data from hospitals and procedures performed
in hospitals are available, so no conclusions can be drawn about the performance or
extent of surgery or the use of RAI. However, it should be noted that thyroidectomy
is still the standard treatment for thyroid cancer in Germany and that patients with
DTC in Germany are often treated postoperatively with ablative RAI therapy. However,
a new German guideline for the treatment of thyroid cancer, which recommends a less
aggressive treatment approach in the context of surgery and RAI therapy, is in
progress. Another limitation of our study is that due to the small number of cases
for some subtypes of the respective lymphatic and hematopoietic tumor diseases, we
could only roughly classify them as lymphoma or leukemia. There is also a lack of
detailed information on some other risk parameters associated with the tumorigenesis
of thyroid cancer, lymphoma, or leukemia that could represent potential confounders
(e. g., external radiation exposure, genetic factors). As a result, our study is
only able to show an association, not a cause-and-effect relationship. Nevertheless,
this study represents an important contribution to the literature, as other authors
have only included patients from hospitals [2323 ] or tumor registries [2121 ]
[2222 ]
[2424 ], and the outpatient setting of general
practitioners and specialists has not yet been studied.
Conclusion
In conclusion, our study provides valuable insights into the complex interplay
between thyroid cancer and hematologic malignancies. The results underscore the
importance of individualized risk assessment and surveillance strategies for thyroid
cancer patients to effectively detect and treat potential SPMs. Further research is
needed to elucidate the mechanisms underlying the observed associations and to
develop optimal treatment strategies for these patients.
