Age greater than 60 years portends a worse prognosis in patients with papillary thyroid cancer: should there be three age categories for staging?

Thyroid carcinoma is a disease of thyroid tissue, and ranges from well-differentiated to poorly- differentiated. The lifetime risk of developing thyroid carcinoma is approximately 1% in the United States, with the well-differentiated subtype representing most of these cases [1]. As is true of thyroid nodules, thyroid carcinoma is 2–3 times more common in women than men [2]. Overall, the incidence of thyroid carcinoma is increasing, and now represents the 5th most common malignancy in women [2‐4].

Age has long been recognized as an important prognostic factor in papillary thyroid cancer (PTC), with better survival observed in patients < 45 years of age, regardless of stage [5‐7]. Given that age is one of the most important prognostic factors in PTC, staging systems utilize an age cut-off of 45 years to stratify risk [8].

Recent studies from Asia and Europe have reported that PTC patients older than 60 years tend to have significant shorter survival time compared with younger patients, even after accounting for other predictors [8‐10].We sought to determine whether age ≥ 60 years is an adverse prognostic factor for disease-specific survival (DSS) and disease-free survival (DFS) in patients with PTC diagnosed and treated in the United States.

This study was approved by both the California Health and Human Services Agency Commission for the Protection of Human Subjects (CPHS), and by the Institutional Review Board of City of Hope Medical Center. Informed consent was waived, since all data is de-identified.

PTC is the most common subtype of thyroid cancer, and age at diagnosis is an important prognostic factor [8, 18, 19]. Age < 45 has historically been used as a cut-off to stratify risk [8]. However, this cut-off makes sense only if one assumes that all patients over the age of 45 have a uniform risk of recurrence and death, whether they are close to the cut-off or far. Recent data has suggested that patients over the age of 45 have progressively worse survival as they age [8‐10]. For this reason, another age cut-off of 60 years has been considered. Studies from Europe and Asia have suggested that this second age cut-off may provide additional prognostic information [8‐10]. Elisei et al. evaluated a large series of patients with differentiated thyroid cancer from a single institution, analyzing overall survival according to age < 45, 45–60 and > 60 years. They demonstrated that age over 60 years, along with the presence of metastatic disease at diagnosis were the two most important prognostic factors for a shorter survival [20]. Our study has shown similar results, but in a large state database, adding to the body of evidence that age > 60 is an important prognostic factor. Although it is possible that there is a “continuum” of the age related to increased risk of death, the differences in survival from one year of age to the next are so small, as to be indistinguishable. In addition, there is no clear inflection point on ROC curve analysis, and we therefore used the cut-off of age > 60, as reported by other studies.

It is somewhat surprising that only ~ 60% of patients received RAI while ~ 90% of patients received surgeries that would result in near removal of the whole thyroid. In the past, the large majority of patients would receive RAI to ablate the small amount of residual tissue resulting from these procedures. This may reflect the trend toward some limitation of the use of RAI in selected cases and indeed the most recent ATA guidelines published in 2009 do provide recommendations for not treating with RAI [21]. While the overlap of this study with these guidelines was only from 2009 to 2011, it is likely that these trends were occurring prior to the publication of these guidelines. When use of RAI over the study period was examined, factors associated with increased use of RAI included age < 60 years, Hispanic or white race, private/HMO insurance, presence of regional (rather than local or distant) disease, operative procedure of total thyroidectomy with cervical lymphadenectomy (likely correlates with stage of disease), presence of multifocal disease, treatment at a high volume center, and treatment at a non-NCI hospital. There was no association between RAI and gender. However, it is also possible that the finding that ~ 60% of patients received RAI can be explained by a coding error. In our cohort, 10% of patients receiving RAI were coded as having undergone lobectomy. Patients undergoing lobectomy alone would not be expected to receive RAI, and it is therefore possible that these patients’ procedures were mis-coded as “lobectomy”, when they had actually undergone a completion lobectomy. Such coding errors are known to occur in large databases [22, 23]. As expected, the use of EBRT was very limited, with only 1.0% of patients receiving this therapy. Although EBRT is used more liberally in parts of Asia [18], it is used only very selectively in the United States.

Our analysis suggests that RAI is associated with significantly improved DSS for the entire cohort (all ages). In light of the 2009 ATA guidelines, RAI would have been used in higher-risk patients with PTC, but not in low-risk patients. Although our 12-year cohort included 9 years of patients diagnosed before publication of the 2009 ATA guidelines, it is likely that clinicians were using clinical judgement in identifying high-risk patients, for whom RAI would be beneficial, even before this recommendation was reflected in national guidelines. It is therefore not surprising that RAI is associated with improved DFS and DSS, given the use of RAI only in those with high risk disease [21].

It is interesting that univariate analysis showed that patients ≥ 60 were more likely to be male, present with tumors ≥ 5 cm, and have metastatic disease at presentation. Because PTC is more common in females, the larger-than-expected percentage of elderly men with PTC in our cohort is somewhat surprising. PTC in males tends to be more aggressive, so our findings that these individuals also tend to have larger tumors and metastatic disease at diagnosis is entirely consistent with accepted principles that outcomes are better in women than men [21]. However, a recent publication evaluated the relationships between gender, age, and PTC outcomes, and found that while younger women tend to have better survival than men, outcomes are similar in older patients [24].This suggests that older age may modify the effect of gender on outcomes from PTC, and that an older age cut-off may be warranted [24].

Additionally, older patients were less like to undergo total or completion thyroidectomy or RAI, and were more likely to receive EBRT. EBRT is indicated primarily for adjuvant use in high-risk patients, or those with unresectable cancer. The higher frequency of EBRT in older patients in our cohort may represent a tendency for surgeons to limit extent of surgical treatment in favor of a lesser surgery in combination with radiation, in patients with co-morbid conditions [25].

We have also shown that DSS and DFS are decreased in patients ≥ 60 years, after controlling for gender, race, co-morbidities, stage, tumor size, extent of surgical treatment, RAI, insurance status, and hospital volume. Similar studies from Asia and Europe have shown that patients ≥ 60 years fare worse than those over 45 years, but younger than 60 years. Our study confirms the importance of age ≥ 60 years on PTC-related survival in an American cohort. This suggests that the biology of disease in elderly patients is inherently more aggressive than disease in younger individuals. Whether this is due to the accumulation of exposures over a lifetime (ionizing radiation from the environment, other environmental toxins, or some other risk factor) is unclear from this analysis. Additionally studies are needed to investigate the underlying causes of the decreased survival in elderly patients with PTC.

The recent publication of the 8th edition of the TNM staging guidelines for thyroid carcinoma changed the “low risk” age cut-off rom 45 years to 55 years. This resulted in many patients being reclassified with low risk disease [26]. Our data would suggest that another “high risk” cut-off of greater than 60 years should be considered.

This study has noteworthy limitations. While the linked data set includes extensive demographic and clinical information on more than 3.4 million cases of cancer diagnosed in California, potentially important covariates were unavailable. Although history of radiation exposure is associated with increased the risk of thyroid malignancy, and with more aggressive forms of disease [27], this information is not collected by CCR or OSHPD. Neither database collects information on previous radiation, and therefore, this was not controlled for in our analysis. Additionally, although insurance status was not a significant predictor of DSS, other socioeconomic factors, such as rural vs. urban living, distance to health care facility, or income may influence timing of seeking medical attention, or availability of endocrinologists and endocrine surgeons. Furthermore, while the data set captures hospital volume, it does not include reliable information on individual surgeon volume, which is known to affect patient outcomes [28]. Coding errors also occur in large, population-based databases, and these would be expected to adversely affect the reliability of the data in regards to RAI and surgical procedure performed, in particular [22, 23]. Patients coded as “lobectomy” may have actually undergone “completion thyroidectomy”, but the veracity of the coded procedure cannot be verified in this dataset, as it is de-identified. While single-institution studies are not subject to issues of coding, as occur in population-based datasets, they also cannot achieve accumulation of similarly large of numbers of patients. Although some studies have suggested that coding errors may occur in up to 10% of records, with sufficient numbers of subjects, this is unlikely to affect study results. However, we acknowledge that this remains an important limitation of studies employing large population-based datasets (such as this one) until better methods of data entry and extraction are available, or until records can be retrospectively verified. Finally, we acknowledge that some of the decrease in DSS and DFS observed in patients ≥ 60 years may be due in part to delay in diagnosis in elderly patients. Additional studies are warranted to further examine this possibility.

The strengths of our study include that information was obtained from a large, prospectively-maintained population-based data set, including all cases of PTC diagnosed and treated in California from 2000 to 2011. Additionally, the linkage between the CCR and the OSHPD data sets, allowed us to control for number of co-morbidities and treatments rendered.

Patients ≥ 60 years of age have worse DSS and DFS after a diagnosis of PTC, across all stages of disease. Given that patients over the age of 45 years have progressively worse survival as they age, these data support having three age groups, 18–44 years of age, 45–59 years, and ≥ 60 as an independent predictor of survival and recurrence to current staging guidelines.