The use of highly sensitive molecular tests to identify recurrent or progressive disease using tissue and/or tumor-specific markers have been used to detect metastases in bone marrow, lymph nodes, peripheral blood, and other sites. Methods employed include RTT-PCR amplification of tissue or tumor-specific transcripts or isolation of cancer cells directly using cell sorting. These approaches are particularly attractive for thyroid cancer because, in comparison to other solid tumors, initial therapy of thyroid cancer frequently results in the removal and ablation of all thyroid tissue, making both tumor and tissue-specific markers useful for early diagnosis. Several markers have been applied to nodes (Table 1) and peripheral blood (Table 2).
The most common sites of tumor metastases in thyroid cancer are local-regional lymph nodes, particularly for papillary cancer. These metastases are frequently present at diagnosis and can be difficult to isolate and eradicate. Standard approaches to diagnosis of local nodes include the level of elevation of serum thyroglobulin concentrations, the presence of abnormally sized or appearing nodes on anatomic imaging often with abnormal cytology on FNA, or iodine uptake in an extrathyroidal location. The diagnosis of metastatic thyroid cancer within a node frequently is confirmed by FNA, but
Table 2. Published diagnostic peripheral blood markers for thyroid cancer
Thyroglobulin Thyroid Peroxidase Ret/PTC Oncogenes Cytokeratin 20 TSH Receptor Human Kallikrein 2
(66-77, 82-85, 87,88) (67, 74, 84) (67) (58) (76) (86)
this method is difficult for small nodes in the neck bed where the amount of aspirated tissue may be small. To enhance diagnostic sensitivity, there has been an interest in developing RT-PCR based approaches to amplify thyroid-specific transcripts from node FNA for both thyroid cancer derived from for papillary and follicular thyroid cancer and for medullary thyroid cancer.
Arturi, et al. (56) reported their experience using RT-PCR amplification of thy-roglobulin and TSH-receptor mRNAs from nodal tissue obtained by FNA of 46 lymph nodes and compared them to cytopathology, thyroglobulin immunoassay of the aspirate fluid, and final histopathology. RT-PCR detected thyroid transcripts in 41 of 41 histopathologically confirmed metastatic tumor samples, including 45% that were inadequate or false negative by standard cytopathology. Similar results were obtained by Gubala, et al. (57) who reported their experience in 70 nodes aspirated from 60 patients with suspected thyroid cancer recurrence. Taken together, these data confirm that thyroid-specific mRNAs can be amplified from nodes in patients with metastatic thyroid cancer, that false positives from ectopic transcription in lymphocytes appears to be uncommon using these particular primers, and the overall accuracy may be adequate for clinical use. Weber, et al. (58, 59) used a slightly different approach, amplifying cytoker-atin 20, an epithelial cell tumor marker, mRNA using RT-PCR from nodes suspected of harboring metastatic differentiated thyroid cancer. In comparison to cytokeratin 20 immunhistochemistry and cytology, the molecular diagnostic approach was more sensitive.
This group has also reported similar data for patients suspected to have recurrent medullary thyroid cancer in cervical nodes. The report that amplification of cytoker-atin 20 and preprogastrin mRNA, a marker of neuroendocrine tumors, by RT-PCR demonstrated enhanced sensitivity and specificity over routine cytology (60, 61). These results, in combination with detection of medullary cancer-related mRNAs in peripheral blood of patients suggest this approach may be useful for patients with medullary cancer (62).
The importance of detecting metastases earlier has not been clarified in thyroid cancer, a disease that typically follows an indolent course. However, for patients with malignant melanoma, amplification of tyrosinase mRNA from sentinel lymph node tissue removed at surgery correlates with development of metastatic melanoma and subsequent prognosis (63, 64). With time, it is likely that early detection will result in better prognosis. The development of markers of aggressiveness, such as p53 mutation analysis, may provide additional predictive data that will help clinicians stratify patients for appropriate treatment paradigms. Other markers derived from cDNA array analysis may also be particularly useful in the future.
The most frequently employed tests for monitoring patients with thyroid cancer for tumor recurrence are measurements of circulating serum thyroglobulin concentrations and radioiodine scanning, both of which rely on thyroid-specific gene transcription or function. Non-thyroid specific monitoring methods include ultrasound, magnetic resonance imaging, computed tomography, positron emissions tomography, and physical examination. Thyroid-specific monitoring, rather than tumor-specific monitoring is particularly useful for patients treated with thyroidectomy and radioio-dine ablative therapy who are, theoretically, devoid of all thyroid tissue, benign or malignant.
The development of more sensitive and specific thyroglobulin assays has led to increased dependence on this test in monitoring paradigms. The ease of a simple blood test and the lack of exposure to radiation are two advantages of this method. However, there are several important limitations of serum thyroglobulin monitoring; 1) circulating autoantibodies directed against thyroglobulin (anti-thyroglobulin antibodies) interfere with clinical assays in approximately 20% of patients, and 2) stimulation of thyroglobulin transcription and release with either endogenous or exogenous thy-rotropin (TSH) is required for adequate clinical sensitivity (65). There has therefore been an interest in developing new assays for thyroid cell detection that are not altered by antibodies and are sensitive enough to not require TSH stimulation.
Ditkoff, et al. (66) reported results from 100 individuals including 87 with thyroid cancer, 6 with benign thyroid disease (nontoxic goiters), and 5 normal subjects following total thyroidectomy (except normal subjects). Total RNA was isolated from the macrophage layer of peripheral blood, and, using RT-PCR amplification of thy-roglobulin mRNA, they detected thyroid transcripts in blood from 9 of 9 patients with metastatic thyroid cancer, but from only 7 of 78 patients thought to be free of disease, and no patients having surgery for benign disease or normal control subjects. Detailed clinical information was not included regarding the clinical status of the patients and TSH levels were not reported. However, these investigators clearly demonstrated that thyroglobulin mRNA could be amplified from peripheral blood and that its presence appeared to correlate with stage of disease.
Tallini, et al. (67) subsequently reported data using different RT-PCR assays for detection of thyroid transcripts from peripheral blood. In this study, the investigators evaluated 44 patients including 24 with thyroid cancer (16 with metastases and 8 free of disease), either pre-operatively, postoperatively, or at both time points for peripheral blood expression of thyroglobulin, thyroid peroxidase, and the RET/PTC1 thyroid oncogene. 56% of the patients with either local or distant metastases had positive assays, compared to 63% of those thought to be free of disease. Of those thought to be free of disease that had positive assays, 80% had cervical adenopathy at diagnosis and were felt to be at high risk of tumor recurrence. Of the patients with benign disease, 2 of 20 patients had a positive mRNA assay, both of which reverted to negative after surgery. The in vitro sensitivities of this assay were approximately 50 cells/ml of blood. Technically, these authors isolated total RNA from whole blood drawn into EDTA-containing tubes and did not isolate a buffy coat layer.
Ringel, et al. (68) also developed a thyroglobulin mRNA assay designed for detection of circulating thyroid cells. The method employed in this study used whole blood placed directly into an RNA-stabilization solution and resulted in a more sensitive assay. In this study, 87 individuals with thyroid cancer were evaluated. Thyroglobulin mRNA was detected in all 14 cervical or distant metastases during L-T4 therapy, while 65% of patients with thyroid bed uptake and 20% of patients with no uptake had detectable thyroglobulin mRNA. These data suggested both a high sensitivity and lower specificity of the assay than the prior studies. Of concern was that similar to the patients with multinodular goiter analyzed by Tallini et al. circulating thyroglobulin mRNA was detectable in all of the normal subjects evaluated and in 20% of athyreotic patients. These results raised the possibility that thyroglobulin may not represent a truly thyroid-specific transcript and that this more sensitive assay detected ectopically transcribed of thyroglobulin in non-thyroid cells. Alternatively, the assay could have been detecting very early minimal residual or recurrent disease.
Additional data have been published from many groups using similar qualitative approaches to amplify thyroglobulin and other mRNA transcripts from peripheral blood. The results have been remarkably variable, with some groups demonstrating excellent correlation between tumor stage and results (69-72), while others demonstrate no correlation with tumor stage (73-75). Several have concluded that the assay is more useful for papillary rather than follicular cancer (69), while others have demonstrated optimal screening by combining thyroglobulin mRNA with new highly sensitive thyroglobulin immunoassays (71). Taken together, nearly all groups have confirmed the presence of circulating thyroglobulin mRNA in peripheral blood of normal subjects, and in a subset of athyreotic patients, suggesting that ectopic transcription of thyroglobulin or splice variants of thyroglobulin can be detected.
The importance of assay methodology has been highlighted in several recent studies. Bojunga, et al. (73) reported data using low and high sensitivity qualitative thyroglob-ulin mRNA assays in patients with thyroid cancer. Using a lower sensitivity assay, they detected circulating thyroglobulin mRNA in 69% of patients with metastatic disease, 46% of patients with thyroid cancer thought to be free of disease, 25% of patients with benign thyroid disease and 18% of control patients. The more sensitive assay increase sensitivity modestly, but resulted in the complete loss of specificity. Gupta, et al. (76) created PCR primers designed to carefully avoid amplification of all known splice variants of thyroglobulin and the TSH receptor. Using these PCR primers, these authors reported detection of thyroid transcripts in 83% of thyroid cancer patients with positive compared to 5% ofpatients with negative radioiodine scans. AH normal volunteers were negative. The specificity was slightly greater for TSH mRNA detection rather than thyroglobulin mRNA detection. Similarly, Savagner, et al. (77) designed thy-roglobulin primers that amplified known splice variants and others that did not. They determined that the splice variants account for approximately 1/3 of the total amplified thyroglobulin mRNA, and that when the primers that do not amplify the region are used, the results correlated with the volume of thyroid tissue and TSH concentration. Taken together, these data clearly demonstrate the importance of methodology in performing these assays, and in proper evaluation of the published data. Differences in sensitivity could be due to the method of sample collection, storage of samples between the phlebotomy and RNA isolation, the specific method for reverse transcription and the PCR primers employed.
Due to the subjective nature of PCR and the apparent discrepancy in the results of studies using qualitative RT-PCR systems, there has been interest in attempting to quantify peripheral blood RT-PCR assays in order to define a clinically relevant level of detection. The advent of real-time quantitative PCR has enabled testing of this approach in clinical trials. Similar to quantitative RT-PCR, the methodological issues are considerable, particularly when attempting to detect very rare transcripts within a particular sample. Other major issues when considering quantitation of RNA is normalization to a control transcript. Traditionally, normalization to glyceraldehyde-3-phosphate-dehydrogenase (GAPDH) or beta actin has been employed; however, tremendous variability in these control transcripts has been reported (72-74). An alternative is normalization to total RNA (18S), while others have chosen not to normalize transcripts at all and normalize to the original blood volume (78-80). This also may not be an accurate method and the use of a "geometric" panel of markers has recently been suggested (81). Thus, it is apparent that normalizing to different control transcripts clearly will alter the reported results and, to date, no standard method has been applied by all laboratories; however, it appears clear that normalizing to a single "housekeeping" gene such as GAPDH or beta actin is likely not appropriate for these samples (81).
Wingo, et al. (82) reported the first quantitative thyroglobulin mRNA assay. In this study, total RNA was derived from peripheral blood samples and the assay was extensively tested. Calibration assays revealed interassay variability of 17-22% due primarily to RNA stability, RNA handing and the reverse transcriptase reaction. The assay displayed reproducible results over a three log concentration range. Ringel, et al. (83) subsequently used this assay to analyze peripheral blood RNA from 107 patients with thyroid cancer; including 84 during L-T4 therapy, 14 following L-T4 withdrawal, and 9 before and after thyroxine withdrawal. Twenty-three patients had circulating anti-thyroglobulin antibodies. Using an arbitrary cut-point to identify patients as either positive or negative for detection (36 PCR cycles), thyroglobulin mRNA measurement assay was more sensitive than thyroglobulin immunoassay, but was less specific at detecting the presence of local and distant metastases. In addition, while there was a statistical correlation between the level of thyroglobulin mRNA and the presence of thyroid tissue on scan, the level of thyroglobulin mRNA did not correlate well with stage of disease. Importantly, the assay appeared to be unaffected by circulating anti-thyroglobulin antibodies, suggesting that perhaps Thyroglobulin mRNA could be used as an adjunctive test to identify patients with recurrent or residual thyroid tissue in the presence of anti-thyroglobulin antibodies. However, the authors cautioned that there was significant overlap between the patients with positive results without definable disease and those with disease, a factor which may limit the usefulness of this particular assay method in clinical practice. Thus, for individual patients, the absolute value of thyroglobulin mRNA did not appear to be diagnostically useful, but the presence or absence of thyroglobulin mRNA might be useful. In addition, similar to other studies, even using a cut-point, a significant minority (38%) of patients with no evidence of disease had positive results and many had detectable values below the cut-point. The relevance of an isolated thyroglobulin mRNA level is uncertain as it might reflect a false positive result from ectopic expression, or the presence of bona fide residual thyroid tissue.
Savagner, et al. (77) developed a quantitative assay for measurement of thyroglobulin mRNA in peripheral blood. In this study, the cut point of a positive or negative assay was determined to be the amount of circulating prostate specific antigen mRNA as a control transcript, no internal normalization was performed and results were reported per total RNA amount. The results in this study were similar to those of Ringel, et al. in that using a mean value, there was a statistical correlation with the absence or presence of residual or recurrent thyroid tissue, but there was significant overlap between all groups for individual data.
Similar to the experience with qualitative thyroglobulin mRNA assays, variable results have also been reported with the quantitative approach. Some ofthese differences are methodological (different primers, use of DNase I, normalization), inherent in the assay method (instability of RNA), while others may be interpretive. Takano, et al. (84) performed a study evaluating thyroglobulin mRNA from peripheral blood and similar to Ringel, et al. identified this transcript in all patients. Unlike the prior study, they were not able to correlate levels with stage of disease. However, in this study, the normalization was performed in a different manner (GAPDH), different PCR primers were utilized, and DNase I treatment was not performed, all different from Ringel, et al. Takano, et al. (84) also report similar data amplifying thyroid peroxidase (TPO) as a tumor marker, results that did not agree with those of Roddiger, et al. (74) who reported a better correlation using TPO mRNA amplification than thyroglobulin mRNA in patients with thyroid cancer. Eszlinger, et al. (85) also did not demonstrate correlation between thyroglobulin mRNA levels and the presence or absence of thyroid tissue. They evaluated several different methods of blood collection and also describe important differences in results depending on the types of tubes used for phlebotomy and the time between the sample collection and RNA isolation. These authors used a new set of primers and normalized to beta actin, factors that distinguish their method from others. To further clarify the importance of recognition of assay differences between groups, Span, et al. (75) used the same thyroglobulin PCR primers as earlier reports and were not able to confirm a relationship between stage of disease and level of thyroglobulin mRNA. However, distinct from those reports, the authors used a different method of RNA isolation and normalize their results to beta actin, both important differences in assay methods that can alter results.
Additional markers, such as cytokeratin 20 and human kallikrein 2 mRNA amplification have recently been reported to have potential diagnostic benefit for thyroid cancer patients (58, 86). These are not thyroid-specific, but may be cancer-specific. These preliminary data require confirmation, but may be an interesting alternative approach to molecular diagnosis of metastatic disease.
Thus, based on these data, it seems that there is clear evidence of ectopic expression ofthyroglobulin, or at least splice variants ofthyroglobulin in non-thyroid tissues. Assay quantitation to "subtract out" this amplification is of uncertain value due to differences in the reported methods and the challenges of normalization of results. Further study and clarification of these issues, in particular, the use of primers that do not amplify splice variants, determination of the best processing protocol for blood RNA isolation, and whether an appropriate form of normalization exists are required before a clear assessment regarding the clinical usefulness of this approach to molecular diagnosis can be made.
Was this article helpful?
Complete Guide to Preventing Skin Cancer. We all know enough to fear the name, just as we do the words tumor and malignant. But apart from that, most of us know very little at all about cancer, especially skin cancer in itself. If I were to ask you to tell me about skin cancer right now, what would you say? Apart from the fact that its a cancer on the skin, that is.