Abstract
However, it is less clear whether the autoimmunity is the cause or effect of the thyroid malignancy. Careful examination of the thyroid cancer and the mainstream oncology literature reveals that either position has its merits (1 –3). On the one hand, oxidative stress and pro-angiogenic effects of autoimmune glandular inflammation might add to the intrinsic oxidative stress within the thyroid gland, thereby promoting chemical mutagenesis and ultimately tumorigenesis and progression (3,4). On the other hand, it seems obvious that tumorigenesis, by its very nature, might lead to the display of hidden or altered antigens, triggering immune responses.
Perhaps both scenarios are occurring, with the situation being complicated by a spectrum of different diagnostic criteria having been used to define thyroid autoimmunity. These have ranged from a history of clinical Hashimoto's or Graves' disease, to serological evidence for thyroid autoantibodies, to histological confirmation of thyroid autoimmunity. Depending on what patient population is studied and which method is used to ascertain the presence or absence of autoimmunity, a study's conclusions might be biased either toward thyroid autoimmunity being a cause or an effect of thyroid cancer. For example, if one examines thyroid glands from patients with Graves' disease who have undergone surgery, one would be looking at a population of patients with very active organ-specific autoimmunity and one might expect to see extremes of oxidative organ stress and neovascularization, as well as a general follicular growth stimulus. Moreover, meticulous histology might reveal microcarcinomas that could have been overlooked by noninvasive methods. This combination of factors would perhaps enrich a study population with cases of apparent causative contributions of autoimmunity. By contrast, at the other end of the spectrum, a cohort of patients who are biochemically euthyroid with no clinical evidence of autoimmune thyroid disease might be expected to contain very few individuals with aggressive thyroid autoimmunity at the histological level, even if some of the patients are serum positive for thyroid autoantibodies. In these cases, the serological thyroid autoimmunity is possibly more likely to represent a host response to the tumor.
In their paper, Kim et al. (1) have studied just such a population; specifically, for all patients who underwent ultrasound-guided fine needle aspiration biopsy (FNAB) of thyroid nodules at their institution over a 3-year period, they retrospectively examined whether the presence or absence of thyroid cancer was correlated with the serum concentrations of thyrotropin (TSH) or autoantibodies against thyroid peroxidase (TPOAb) or thyroglobulin (TgAb). They also examined whether histological evidence of autoimmunity predicted the presence of thyroid carcinomas in a subset of patients who underwent surgery. They excluded patients from the study who were not biochemically euthyroid at presentation as well as those who received antithyroid drugs or thyroxine treatment. This approach effectively eliminated most, if not all patients with significant preexisting autoimmune thyroid disease from the cohort.
For this preselected group, their study answers the question of cause or effect pretty clearly. Thyroid autoimmunity is a consequence of thyroid cancer in these patients, not its cause, which is consistent with the considerations outlined above. While this is a worthwhile observation in its own right, the most interesting finding of the paper is that only seropositivity for TgAbs, but not for TPOAbs, was associated with the presence of thyroid cancer. This is in contrast to other studies that have not preselected their patients for the absence of overt autoimmune thyroid disease and have generally found an equal or higher prevalence of serum TPOAbs compared to serum TgAbs (5).
It is possible that the higher relative abundance of TgAb in thyroid cancer patients in the present study is an artifact of the retrospective study design—more patients had measurements of TPOAb than TgAb—or that it is related to the analytical methodology—the TgAb and TPOAb assays might have had different detection sensitivities. However, the authors' speculation that thyroglobulin (Tg) in thyroid cancer patients might differ from ordinary Tg, potentially rendering it more immunogenic, is supported by several previous publications since the 1980s through to the more recent past (6,7). Whether such increased Tg immunogenicity would translate into different epitope recognition patterns remains unclear (8,9); it might simply increase the likelihood of TgAb being formed. It would be fascinating to study such antibodies in more detail, as recent investigations have shown that tumor-specific antibodies seem to belong predominately to the IgM class and may be largely germ-line encoded rather than somatically rearranged, suggesting they belong to the natural, innate immune system rather than the realm of adaptive/acquired immunity (10). In this context it is interesting to speculate whether Tg's heavy glycosylation and its potential post-translational alteration within tumor cells play a role in this antibody response; many tumor-specific autoantibodies seem to recognize preferentially abnormal or new carbohydrates on tumor-derived proteins (10).
On a practical level, the association of serum TgAb with the presence of malignancy in a thyroid nodule suggests that TgAb could serve a role, albeit an ancillary one, as a diagnostic marker. While the increases in relative cancer risk observed in TgAb-positive patients in the present study are not large, further characterization of the antibodies might improve the predictive value; for example, by showing that they are IgM. Used in conjunction with other clinical, laboratory, and imaging indicators, this could improve prebiopsy prediction of malignancy in thyroid nodules significantly. In this context, the authors' other observation, an almost linear increase in malignancy risk with rising serum TSH concentrations, is also noteworthy. While their use of parametric statistical tests for comparisons of TSH levels between groups is debatable—log transformation before analysis or nonparametric methods would seem more appropriate, given that TSH values are usually log-normal distributed—the finding is interesting and provides potential proof of principle that multimarker combinations could indeed improve prebiopsy prediction of thyroid malignancy. What needs to be done ultimately, however, is to show that such an approach could potentially avoid FNABs (e.g., for tiny nodules with favorable multimarker profile) or aid in deciding on further management or testing in cases of nondiagnostic or inconclusive FNABs.