Introduction

A progressive increase in the incidence of thyroid cancer (TC) has been reported in various countries over the last few decades [1]–[3].

The hypotheses to explain these epidemiological data include the increasing use of procedures with progressively increasing sensitivity for diagnosing TC, as well as an increased exposure to environmental factors, such as ionizing radiation [1]–[4]. In the former case, compared with palpation alone on physical examination, the increased detection of TC within a thyroid nodule (TN) has been facilitated with the use of needle aspiration techniques (NAT; including fine-needle aspiration [FNA] for cytological examination; large-needle aspiration biopsy [LNAB] for histological examination and, if available, molecular analyses; and ultrasound [US]-guidance) [1]–[7].

Therefore, the question arises whether the increased TC incidence is due to the improved NAT or rather, some increased or more potent effect of carcinogenic factors specific for malignant, but not benign TN [1]–[3]. This retrospective study addresses this issue by analyzing the proportion of TN harboring a TC during a four-decade period at the University Hospital of Pisa.

Methods

Results

The clinical data of the 3 TN patient groups are given in Table 1. Patients were primarily from the central and southern regions of Italy.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Incidence of thyroid cancer in patients with palpable thyroid nodules submitted to or selected for surgery^*.

https://doi.org/10.1371/journal.pone.0047758.t001

From 1972 to 2010, 1,766 TN patients were operated on following two different selection criteria. A total number of 231 cancers were found at postoperative histology. This figure corresponds to 2.85% of all the preoperatively selected TN patients. The distribution of these thyroid cancers in the 3 periods considered was the following: 1972–1979, 35 cancers (papillary 21, follicular 11, medullary 2, and anaplastic 1) corresponding to 3.07% (95% CI: 2.15–4.24%) of all the preoperatively selected TN patients; 1980–1992, 150 cancers (papillary 120, follicular 23, medullary 3, anaplastic 3, and squamous 1) corresponding to 2.77% (95% CI: 2.35–3.25%) of all the preoperatively selected TN patients; 1993–2010, 46 cancers (papillary 41, follicular 4, anaplastic 1) corresponding to 2.93% (95% CI: 2.16–3.89%) of all the preoperatively selected TN patients. The percentage of TC was not significantly different in the 3 considered periods (Fisher-Freeman-Halton's exact test p = 0.810).

A significant change in the cancer histotype composition during the 3 considered time periods was observed. The papillary histotype significantly increased (p = 0.008) while the follicular histotype significantly decreased (p = 0.027).

Discussion

This study shows that the proportional rate of TC among patients with a TN examined by the same physician within the University Hospital of Pisa and morphologically investigated at the same Institute of Pathology did not significantly change in the last 40 years. This consistency occurred even though the methods of cancer detection changed from minimal preoperative selection with extensive surgery to more aggressive selection by sensitive NAT. This observation is corroborated by the data from a different 34,266 TN patient series evaluated in a different Department of the University Hospital of Pisa from 1997 to 2004 [13]. In this study, there were 3,406 excised thyroid nodules with 1,208 thyroid cancers in 3,004 patients, of which 1062 had TC (3.1%). This proportional rate is very similar to that observed among our 3 study groups.

These two clinical series represent the principal source of data on the TC proportional rate among patients with TN within the University Hospital of Pisa. Our study first monitors the proportion of TC among TN in a consecutive long series of data from the same team. All the other studies are substantially cross-sectional and do not report any information on the time trend. During the study no significant change in the proportional rate of TC among TN patients occurred. This finding suggests that pathogenetic risk factors and more sensitive diagnostic technologies did not differentially affect the incidence of TC and TN.

In 2000, the TC incidence among European countries varied by a factor of 3–4 (from 2–3 to 8–9 per 100.000 women) with progressive increases in some countries (e.g., Finland and France) and no changes in others (Sweden and Norway) [1]. In the United States, Davies and Welch [2] reported that the incidence of TC more than doubled over the past 30 years (from 3.6 per 100.000 in 1973 to 8.7 per 100.000 in 2002) and that 87% of the increase was due to the diagnosis of small papillary cancers (≤1 cm). In Italy, Buzzoni [3] reported that from 1993–1995 to 2003–2005, the crude TC incidence rate increased 118% (from 6.5 to 14.2 cases for 100.000 inhabitants/years), and that the increase due to an aging population was only 6%. Similar TC incidence data in Italy were reported by Dal Maso et al. [14] when comparing the periods 1991–1995 and 2001–2005. Different TC incidence rates with similar selection protocols and technologies, or varying ages and study periods, raised the question of a changing carcinogenic environment [15].

Risk factors for TC [1] include exposure to ionizing radiation in childhood, family history, other thyroid diseases, dietary iodine, and environmental pollutants (endocrine disruptors). Only relatively high doses of ionizing radiations (>200 mSv) can significantly increase the TC incidence rate [1].

The presence of goiter, TN, hypothyroidism with elevated TSH, or autoimmune disease has been considered as an important risk for TC [16], [17]. Dietary iodine content is a factor affecting the occurrence of goiter or a TN and may therefore have indirect effects. Endocrine disruptors, such as iodine, polychlorobiphenyls, and some pesticides are suspected to be carcinogenetic for the human thyroid, though no conclusive evidence has been demonstrated [1].

Quantitative analysis has shown that, apart from childhood irradiation, a benign thyroid lesion is the strongest risk factor (RR 12–33) for TC [16], [17]. This suggests that cellular proliferation can be a mechanism common to many tumor processes as induced by radiation, TSH, low iodine [13]. It should be noted that any risk factor or process common to both benign and malignant TN may not influence the proportional changes between the two. This is a potential confounder to any conclusion that environmental factors have remained stable over the 40 year study period described herein.

Since the 1950 report of Duffly and Fitzgerald [18] there has been increasing awareness that radiation therapy to the upper body administered to infants, children, and even young adults may induce carcinoma of the thyroid 10 to 30 years later [1]. There was general agreement that the incidence of nodular thyroid disease was much higher in irradiated patients as compared with nonirradiated controls [18]. There was also general agreement that the percentage of malignant TN was similar in both irradiated and nonirradiated groups [19]. Thus, the aim of screening radiation-exposed individuals was to identify those individuals who were at high risk for developing nodular thyroid disease [19]. Our data are consistent with these conclusions that improved detection of any TN, as the strongest risk factor for TC, may be a more important contribution to the TC incidence than diagnostic procedures that identify high risk TN.

When the population exposed to atomic weapons testing in the Marshall Islands were examined approximately 35 years after, the proportional rate of TC among TN was 3.1% [20]. In this population, the proportion of TN patients detected with US was 21.2%. Our data are also consistent with this TC proportion [20].

Following the Chernobyl accident, highly irradiated (10.8 cGy) adults within the first 5 years showed a TN incidence of 1% using US [21]. Four years after the accident, Mettler et al. [22] reported that the prevalence and characteristics of TN in contaminated settlements and unexposed populations were the same (using US screening). In contrast, Drozd et al. [23] reported that among 1,132 highly irradiated children (4–14 years old when exposed and 8–18 when investigated), the proportion of TN patients and the proportion of TC among TN patients were 1.2% and 50%, respectively, much higher than in the control unexposed population where a few benign nodules and no cancer were observed. In 2 studies, each comparing relatively contaminated areas far from Chernobyl, with a control group in more distant uncontaminated areas, no difference was found in children or adolescents, with a ratio of TN subjects/investigated subjects varying from 2% to 6% [24], [25]. Among children and adolescents exposed (0.79 grays) to ¹³¹I after Chernobyl and evaluated 12 to 14 years later, TN was found in 2.7%, with 12% harboring TC [26]. Reiners reported a progressive increase of TC incidence in irradiated children and adolescents up to 10 or 15 years after the accident; however he did not describe TN incidence [27]. These studies suggest that age at the time of the investigations significantly affected TN incidence [22], [23]. However, although the general opinion is that ionizing radiation increases the number of both benign and malignant nodules [19], [20], [22], the Chernobyl data do not clarify whether at some time interval from the irradiation and at some young age, TC incidence rates increase more that those for benign nodules.

A noteworthy finding is that the proportion of follicular cancer was reduced from 31.4% to 8.7% during the study. Possible explanations for this finding are the following. At the beginning of the study some follicular variants of papillary cancers and some largely follicular differentiated cancers were diagnosed as follicular cancer while by the time these diagnoses changed into papillary cancer [5]. The iodine prophylaxis, which was more and more advised and promoted during the study, may have contributed to reducing follicular cancer [28], [29].

Molecular genetics has recently improved the understanding of TC pathogenesis [30] and may represent a useful investigative tool.

Acknowledgments

The authors thank SM Cammisuli for his technical assistance.