- Email: [email protected]
Radioactive iodine offers survival improvement in patients with follicular carcinoma of the thyroid
Radioactive iodine offers survival improvement in patients with follicular carcinoma of the thyroid Yale D. Podnos, MD, MPH, David Smith, PhD, Lawrence D. Wagman, MD, and Joshua D. I. Ellenhorn, MD, Duarte, Calif
Background. The use of radioactive iodine (RAI) in patients with follicular thyroid carcinoma is well established. How its use affects patient outcome and which patients benefit is understood poorly. This study seeks to determine how RAI influences survival and to delineate the populations that are impacted most. Methods. The Surveillance, Epidemiology, and End Results database is a sample of approximately 14% of the US population. It was used to identify patients with follicular thyroid carcinomas and the treatment that they received. Factors such as the presence of lymph node and distant metastases, age, and tumor size were included for analysis. Results. A total of 4317 patients were identified with follicular thyroid carcinoma. Of these, the records of 2112 patients who were entered in the study after 1988 contained the required data and were included for analysis. Median follow-up time was 95 months. Factors that were associated with increased risk of death included distant metastatic disease, cervical lymph node disease, and the lack of RAI use. Protective factors were tumor size of <2 cm and age of <45 years. Some patients with a greater number of risk factors benefited from RAI. Conclusion. RAI provides survival benefit to some patients with follicular carcinoma of the thyroid. The greatest improvements were seen in those patients with locoregional or distant disease spread. (Surgery 2005;138:1072-7.) From the Departments of General Oncologic Surgery and Biostatistics, City of Hope National Medical Center, Duarte, Calif
FOLLICULAR CARCINOMA OF THE THYROID accounts for approximately 15% of the 25,690 cases of thyroid cancers annually.1 In most analyses, it is grouped with the more common papillary type of thyroid cancer. As with papillary cancer, risk stratification schemes with patient age, grade of neoplasm, presence of metastatic disease, tumor size, extent of disease, and completeness of resection often are used to determine a patient’s risk of morbidity and death from follicular carcinoma, with the purpose of helping to guide operative and medical therapies.2-4 Survival after treatment for follicular carcinoma of the thyroid, although not superior
to that of papillary cancer, is generally very good, with 20-year overall survivals exceeding essentially all other visceral malignancies. Because it is most often studied in conjunction with papillary carcinoma, less is known about follicular cancer and its biologic behaviors and response to various treatment regimens. Specifically, controversies exist regarding the optimal extent of surgical resection and the role of postoperative radioactive iodine (RAI). This study sought to determine whether the use of RAI improved overall survival in patients with follicular carcinoma of the thyroid. Additionally, patients were assessed by risk to determine which, if any, groups benefit from RAI.
Accepted for publication September 3, 2005.
MATERIAL AND METHODS The Surveillance, Epidemiology, and End Results (SEER) database is a National Cancer Institute-designed and --maintained catalog of patient data. It is a representative sample of the United States urban population, encompassing approximately 14% of the total population that resides in Atlanta, Detroit, San Francisco, Seattle, Los Angeles, San Jose, Connecticut, Hawaii, Iowa, New Mexico, and
Presented at the 26th Annual Meeting of the American Association of Endocrine Surgeons, Cancun, Mexico, April, 2005. Reprint requests: Joshua D. I. Ellenhorn, MD, Department of General Oncologic Surgery, City of Hope National Medical Center, 1500 E Duarte Rd, Duarte, CA 91010 E-mail: [email protected]. 0039-6060/$ - see front matter Ó 2005 Mosby, Inc. All rights reserved. doi:10.1016/j.surg.2005.09.021
1072 SURGERY
Podnos et al 1073
Surgery Volume 138, Number 6
Fig 1. Overall survival of all follicular patients (P = .104).
Table I. Multivariate analysis Lower Upper Risk ratio CI CI P value Metastatic disease Lymph node disease Lack of radioisotope Tumor size <2 cm Age <45 y
2.47 1.94 1.24 0.72 0.34
1.80 1.47 1.03 0.52 0.24
3.31 2.50 1.51 0.96 0.46
<.0001 <.0001 .026 .022 <.0001
Utah. The SEER database was queried for those patients with a diagnosis of follicular thyroid carcinoma with the International Classification of Diseases for Oncology, ninth revision (C73.9). Patients with histologic subtypes of follicular thyroid carcinoma coded 8330, 8331, and 8332 are included in statistical analysis. No Hu¨rthle cell or follicular variants of papillary cancer are included in the analysis. Patient data from 1973 to 2001 was searched; however, only the data for patients that were entered during or after 1988 contain the necessary information and are included for comparison. Specific data gleaned from SEER with the use of SEERStat included age, primary tumor size, the presence of lymph node metastasis, the presence of distant metastasis, the operation that was performed, the medical therapies that were used, and gender. Data regarding operative treatment, tumor size, disease spread, and specific use of RAI were available only after 1988. Information regarding the recurrence of neoplasm, the chronology of the procedures that were performed, neoplastic invasion, and RAI response were not available in the database. Statistical analyses were performed with (SAS, Cary, NC). Statistical significance was defined as a probability value of <.05. Data were plotted using (Insightful, Seattle, WA).
Fig 2. Survival of patients with distant metastases (P = .028).
RESULTS A total of 4317 patients with follicular thyroid carcinoma were identified from the SEER database from 1973 to 2001. Of these, 2112 patients (1085 patients received RAI; 1027 patients received no radiation) were in the group that were diagnosed in 1988 or later and are used for analysis. The median follow-up time for this group was 95 months. The 14-year overall survival rate for patients with follicular cancer was 80% for patients who received RAI compared with 71% for patients who did not receive RAI (P = .104). Overall survival curves are shown in Fig 1. Multivariate analysis was then performed to determine which factors improved or adversely affected overall survival. The lack of RAI was associated with a risk ratio of 1.24 (95% CI, 1.031.51; P = .026). Distant metastatic disease, the greatest predictor of poor overall survival, was associated with a risk ratio of 2.47 (95% CI 1.80-3.31; P < .0001); disease that had spread to the cervical lymph nodes was associated with a relative risk of 1.94 (95% CI, 1.47-2.50; P < .0001). These 2 factors were associated with one another (P < .0001). Factors associated with improved risk included primary tumor size <2 cm (relative risk, 0.72; 95% CI, 0.52-0.97; P = .022) and age <45 years (relative risk, 0.34; 95% CI, 0.24-0.46; P < .0001). These are summarized in Table I. For each of the aforementioned factors that were found on multivariate analysis to affect significantly overall survival at 14 years, comparisons were made regarding RAI treatment and survival. Treatment with RAI in patients with distant metastases more than doubled overall survival from 17.6% to 38.3% (P = .036; Fig 2). For patients without metastatic disease, a smaller increase in survival from 71.6% to 80.5% was seen with RAI therapy (P = .036). Of
1074 Podnos et al
Surgery December 2005
Table II. Overall survival by risk group Age <45 y Age >45 y Lymph node (ÿ) Lymph node (+) Metastases (ÿ) Metastases (+) Size <2 cm Size >2 cm
Fig 3. Overall survival of patients whose results were lymph node positive (P = .018).
the 76 patients with cervical lymph node disease, the 49 patients who received RAI had a 14-year survival of 55%, compared with 0% for the 27 patients who were treated without RAI (P < .02; Fig 3). Treatment with RAI improved survival in patients with tumor size <2 cm from 87.5% to 96.8% (P = .041). Independently, age >45 years, age <45 years, nor tumor size >2 cm were associated with improved overall survival after RAI. However, large tumor size was associated with older age (P < .05). The full results are summarized in Table II. In each analysis based on size, age, presence of cervical lymph node disease, and presence of distant metastases, several groups were found to benefit from RAI. The most dramatic survival difference occurred in patients with positive lymph node disease, distant metastases, tumors >2 cm, and age >45 years. In this group of only 11 patients, median survival in the 8 patients after RAI was 57 months, compared with only 1 month in the 3 patients who did not receive RAI (P < .01). Patients with distant metastatic disease and tumors >2 cm also benefited from RAI. In this group of 42 patients, the 10 patients who did not receive RAI had a median survival of 17 months, compared with 57 months in the 32 patients who received RAI (P < .01). In an assessment of how the extent of operation affected survival, each permutation of size < or >2 cm, age > or <45 years, the presence or absence of cervical lymph node disease, and the presence or absence of distant metastasis were examined. The only subgroup that derived a survival benefit was patients with all of the following characteristics: tumor size >2 cm, age >45 years, and no distant metastases. In that group of 49 patients, 44 patients underwent total thyroidectomy, and 5 patients underwent a less aggressive resection. Median survival
+RAI (%)
No RAI (%)
P value
95.5 63.5 81.9 54.8 80.5 38.3 96.8 76.6
93.7 53.6 72.5 0 71.6 17.6 87.5 68.3
.959 .233 .588 .018 .036 .028 .041 .213
in those patients who underwent total thyroidectomy was 57 months, compared with 32 months with less thorough resection (P < .04 by Wilcoxon test; P = .09 by log-rank). DISCUSSION Together with papillary and Hu¨rthle cell carcinoma, follicular carcinomas of the thyroid are referred to commonly as well-differentiated thyroid cancer. Follicular carcinoma accounts for nearly 15% of all thyroid cancers, although papillary and its variants comprise approximately 80% of cases. Risk stratifications systems (AMES, AGES, MACIS) were developed with the use of data sets that included both of these major histologic categories in a single group.2-4 However, various thyroid histologic conditions have distinct biologic behavior. It is known that papillary cancers most often spread through the lymphatic system,5,6 whereas follicular carcinoma has a predilection for hematogenous spread.7 Despite the fact that both papillary and follicular cancer share older age, distant metastases, cervical lymph node disease, and large tumors as poor prognostic factors, the difference in overall survival both with and without RAI underscores the different biologic composition of these tumors. In grouping the 2 diseases together, the resulting conclusions overwhelmingly under-represent follicular variants and render the conclusions difficult to extrapolate to the treatment of patients with follicular cancers. Thyroid carcinoma survival, although often excellent, differs greatly between papillary and follicular types. In a large singleinstitution review, survival for papillary cancer at 10 and 20 years was 96% and 95%, respectively.3 For follicular cancer, 10- and 20-year overall survival was 87% and 71%, respectively. Patients with follicular carcinoma in the lower risk stratifications had 20-year survival rates of 86%, although high-risk patients had only 8% 20-year survival. These patients died of metastases to the lungs, bones, brain,
Surgery Volume 138, Number 6
spinal cord, and neck. In that study, cervical lymph node disease, although uncommon, was associated with a worse prognosis, because 60% of these patients also had distant metastatic disease. Our data confirmed this finding and could explain the reasons that these patients with cervical lymph node disease endured a worse overall survival. The use of RAI in patients with well-differentiated thyroid cancer is well-established in the literature. Because of its predominance, the findings of these studies are more applicable to papillary than follicular carcinoma. Even when pooled as a meta-analysis, the benefit of RAI in low-risk patients could not be verified definitively.8 In the present study, RAI appeared to provide a survival advantage when analyzed in a multivariate model. However, the survival benefit of RAI is not uniform. Patients in higher risk groups (the presence of lymph node or distant metastases) derive a survival advantage from RAI, although patients with larger tumors or older age do not. These latter 2 traits are known risk factors in follicular cancer and often are associated with one another. The diminished benefit of RAI with increasing age may relate to the finding that, in older patients (and those with larger tumors), uptake of RAI is less predictable than in younger patients.9 This may explain the reason that, contrary to expectation, older patients and patients with larger tumors are not helped significantly by RAI. As with papillary carcinoma, the adjuvant use of RAI in patients with follicular cancers has never been proved in a prospective randomized trial. In a retrospective review of 135 patients with a median follow-up time of 10.8 years, postoperative RAI in patients with follicular cancer did not reduce the risk of distant recurrence.10 Samaan et al11 found a significant decrease in recurrence but no change in mortality rate or disease-free interval in the 108 patients in their study with follicular carcinoma. In the SEER database of 2112 patients who were entered since 1988, overall survival was improved with RAI by multivariate analysis. The survival benefit extended across some risk stratification groups. By multivariate analysis on the entire data set, these data show that the failure to use adjuvant RAI is associated with a 24% increase in disease-specific mortality rates. On univariate analysis, a trend toward statistical significance existed. Inspection of the graph seen in Fig 1 shows the curves to begin divergence at approximately 10 years. This is not surprising, given the fact that follicular cancer generally is slow growing. Patients who did not receive RAI or those for whom RAI will not result in a survival advantage are more likely die of the disease after approximately 10 years. It is likely that the curves
Podnos et al 1075
will continue to separate, rendering the graph statistically significant in time. Although substantial, it is possible that the survival benefit of RAI was missed by the analysis of previous data sets that were smaller than SEER. Because the SEER database does not include recurrence, we were unable to address how RAI affected recurrence. We can hypothesize, however, that RAI would eradicate remnant tumor nests that lead to a reduction in tumor recurrence. However, data regarding the uptake of RAI and how it affected patient treatment are not included in the SEER database. Clearly, this question is not answered definitively, and additional studies of the effect of adjuvant RAI on tumor recurrence are necessary. The extent of operation required for patients with follicular carcinoma remains somewhat controversial. Some surgeons perform total thyroidectomy on these patients with the sole purpose of being able to provide adjuvant RAI or monitor thyroglobulin postoperatively as a tumor marker.12 Other surgeons advocate total thyroidectomy because these patients with follicular cancers already are placed in the intermediate or high-risk category (along with papillary cancers that have some poor prognostic variables).13 However, a number of studies found no difference in overall survival in patients with follicular cancer by the type of operation that was performed.7,14 In the evaluation of the entire SEER database, multivariate analysis does not reveal any statistically significant amelioration of overall survival on the basis of the operation that was performed. Assessment of groups on the basis of risk factors revealed that the only patients who derived survival benefit from total thyroidectomy were those with the full constellation of older age, large tumors, and distant metastases. Although the log-rank result was not significant (P = .09), statistical significance was achieved with the Wilcoxon test (P = .04). The latter test weighs events that occur early in the analysis more than those events that occur subsequently. As a result, this analysis shows that, despite no overall significant improvement in survival, total thyroidectomy was associated with a reduction in earlier deaths. Despite the recommendation that patients with follicular carcinoma undergo total thyroidectomy, this does not appear to be practiced universally. Haigh et al15 found that patients with follicular thyroid cancer were actually less likely to undergo total thyroidectomy. Additionally, contrary to every well-known risk stratification system, older patients were found to be less likely to undergo more extensive thyroid resections. This paradoxic finding
1076 Podnos et al
might relate to controversy about the extent of operation and adjuvant role of RAI. In clarifying the utility of RAI in most patients with follicular thyroid carcinoma, our results should help alleviate some of the confusion and lack of uniformity with respect to how these patients are treated. With the many considerations that are entertained when treating patients with follicular carcinoma of the thyroid, the decision about which therapies should be used still requires attention to an individual’s risk-factor profile. REFERENCES 1. Jemal A, Murray T, Ward E, Samuels A, Tiwari RC, Ghafoor A, et al. Cancer statistics, 2005. CA Cancer J Clin 2005;55: 10-30. 2. Cady B, Sedgwick C, Meissner WA, Wool MS, Salzman FA, Werber J. Risk factor analysis in differentiated thyroid cancer. Cancer 1979;43:810-20. 3. Hay ID, Grant CS, Taylor WF, McConahey WM. Ipsilateral lobectomy versus bilateral lobar resection in papillary thyroid carcinoma: a retrospective analysis of surgical outcome using a novel prognostic scoring system. Surgery 1987;102: 1088-92. 4. Hay ID, Bergstralh EJ, Goellner JR, Ebersold JR, Grant CS. Predicting outcome in papillary thyroid carcinoma: development of a reliable prognostic scoring system in a cohort of 1779 patients surgically treated at one institution during 1940 through 1989. Surgery 1993;114:1050-8. 5. Shah JP. Differentiated thyroid cancer. In: Bloom HJG, Hanham IWF, Shaw HJ, editors. Head and neck oncology. New York: Ravens Press; 1986. p. 207-14. 6. Attie JN, Khafif RA, Steckler RM. Elective lymph node dissection in papillary carcinoma of the thyroid. Am J Surg 1971;122:464-71. 7. Brennan MD, Bergstralh EJ, van Heerden J, McConahey WM. Follicular thyroid cancer treated at the Mayo Clinic 1946 through 1970: initial manifestations, pathologic findings, therapy, and outcome. Mayo Clin Proc 1991;66:11-22. 8. Sawka AM, Thephamongkhol K, Brouwers M, Thabane L, Browman G, Gerstein HC. A systematic review and metaanalysis of the effectiveness of radioactive iodine remnant ablation for well-differentiated thyroid cancer. J Clin Endocrinol Metab 2004;89:3668-76. 9. Apostoaei AI, Miller LF. Uncertainties in dose coefficients from ingestion of 131I, 137Cs, and 90Sr. Health Phys 2004; 86:460-82. 10. Chow SM, Law SC, Mendenhall WM, Au SK, Yau S, Yuen KT, et al. Follicular thyroid carcinoma: prognostic factors and the role of radioiodine. Cancer 2002;95:488-98. 11. Samaan NA, Maheshwari YK, Nader S, Hill CS Jr, Schultz PN, Haynie TP, et al. Impact of therapy for differentiated carcinoma of the thyroid: an analysis of 706 cases. J Clin Endocrinol Metab 1983;56:1131-8. 12. Duren M, Siperstein AE, Shen W, Duh QY, Morita E, Clark OH. Value of stimulated serum thyroglobulin levels for detecting persistent or recurrent differentiated thyroid cancer in high- and low-risk patients. Surgery 1999;126:13-9. 13. Shaha AR, Loree TR, Shah JP. Intermediate-risk group for differentiated carcinoma of thyroid. Surgery 1994;116: 1036-41. 14. Wanebo H, Coburn M, Teates D, Cole B. Total thyroidectomy does not enhance disease control or survival even in high-
Surgery December 2005
risk patients with differentiated thyroid cancer. Ann Surg 1998;227:912-21. 15. Haigh PI, Urbach DR, Rotstein LE. AMES prognostic index and extent of thyroidectomy for well-differentiated thyroid cancer in the United States. Surgery 2004;136:609-16.
DISCUSSION Dr Subhash Patel (Chicago, Ill). This is a very interesting paper. However, it is limited by the retrospective nature, especially when it comes to lymph node status. Routine lymph node dissection is still controversial and is not performed by most surgeons. Therefore, any analysis or conclusions that are drawn from incomplete data to determine the importance of RAI therapy in relation to lymph node status would be erroneous. Dr Podnos. The problem with thyroid cancer is that we will never get any prospective randomized studies to really answer these questions well. Because of the long nature of the course of the disease, it makes any prospective studies unfeasible. So we must use databases like SEER and single-institutional studies the best that we can. Dr Patel. The lymph node status remains unclear. Lymph node dissection probably was performed in selected patients in whom either they were clinically palpable or appeared large at the time of surgery. What about those patients in whom the lymph nodes are not removed but who may have microscopic metastasis? The survival benefit of RAI therapy in these patients could not be analyzed in this retrospective review. Dr Podnos. That is a good question. Because this is a large database, I cannot account for what happened to each of these patients. It is likely that those patients who were clinically positive regarding lymph node disease but only microscopically positive were in the database as being negative, and nothing has been done about them. That is an interesting point, because that is where the use of RAI in theory could come to help. In papillary cancer, we presented a paper at the SSO that showed that RAI really did not help very much. In follicular cancer, it might help some people and not others. Dr Peter Angelos (Chicago, Ill). In your analysis of the data, can you assess the survival advantage to the use of RAI versus the extent of the operation? In other words, are you comparing total thyroidectomy with RAI to total thyroidectomy without RAI or are you comparing total thyroidectomy with RAI to lobectomy? Dr Podnos. We ran this analysis also based on the operation that was performed, and the operation that was performed did not make a difference, except in one very small, specific group of patients who I did not present here but who are in the article. Those patients underwent a total thyroidectomy and no RAI, and there was no difference. We think that these results are most likely a function of the RAI rather than the operation that was performed. Dr Keith S. Heller (Lake Success, NY). The data are the data, and you have to analyze what you have. Are you concerned that perhaps the SEER database, because of the way it is collected, underreports the use of a treatment such as RAI? I ask because RAI frequently is not
Surgery Volume 138, Number 6
given in the hospital, and most of the SEER data comes from hospital records. Looking at your numbers, specifically for patients with distant metastases, it is difficult for me to believe that there are that many patients with metastatic follicular cancer who did not receive RAI. I suspect it is not picked up in the database. If that group is underreporting the use of RAI, I would be concerned that all your other groups are underreporting the use of RAI as well. Dr Podnos. That, too, was a big concern. That is the reason that, with the last slide, I tried to mention some of the problems with SEER. There definitely are some problems with it, and they need to be understood when looking at the data. There could be decreases in reporting of RAI. In looking at those graphs, if the groups were really the same, I would expect there not to be any differences. The differences in patients with metastatic disease were really quite dramatic. I would tend to believe them. Dr Ashok R. Shaha (New York, NY). It is difficult to look at the data from SEER and not be critical about it. The numbers are large. There has been a flurry of papers in the recent literature from SEER because this is a public domain. You can go into the website and get all the information and try to analyze, reanalyze, and reanalyze. I think we must be very cautious when we make important conclusions based on these data. One of the important factors in follicular thyroid cancer is capsular invasion; that information is not
Podnos et al 1077
available. Anytime we interpret the data based on what we have, we may be drawing some incorrect conclusions. This is a good demographic paper, but I think drawing conclusions would be very difficult. Based on your last slide, I think you may want to conclude that RAI offers improvement in patients with high-risk follicular carcinoma of the thyroid, which that may be easier to conclude than the title as it stands right now. Dr Podnos. As I said, I agree that there are some problems with SEER and that these must be understood in interpreting the data. In terms of capsular invasion, that is something that has been well-established as being a risk factor. It is not in SEER. There are other things that we mentioned that were not overlooked that are in SEER and must be confirmed in subsequent papers. For that type of thing, SEER is very beneficial. Dr Henning Dralle (Halle, Germany). I am a little concerned about the follow-up. I have a very short question regarding the new UICC classification. Did you observe any difference in the group <2 cm with regard to survival or recurrence rates? Dr Podnos. We cannot speak to recurrence because that was not in SEER. We broke it down by size > or <2 cm because that is what TNM states. When size was run as a continuous variable, the results were the same. Dr Dralle. But there are T categories 1-A and 1-B, so you can perhaps analyze these groups as well.
Background. The use of radioactive iodine (RAI) in patients with follicular thyroid carcinoma is well established. How its use affects patient outcome and which patients benefit is understood poorly. This study seeks to determine how RAI influences survival and to delineate the populations that are impacted most. Methods. The Surveillance, Epidemiology, and End Results database is a sample of approximately 14% of the US population. It was used to identify patients with follicular thyroid carcinomas and the treatment that they received. Factors such as the presence of lymph node and distant metastases, age, and tumor size were included for analysis. Results. A total of 4317 patients were identified with follicular thyroid carcinoma. Of these, the records of 2112 patients who were entered in the study after 1988 contained the required data and were included for analysis. Median follow-up time was 95 months. Factors that were associated with increased risk of death included distant metastatic disease, cervical lymph node disease, and the lack of RAI use. Protective factors were tumor size of <2 cm and age of <45 years. Some patients with a greater number of risk factors benefited from RAI. Conclusion. RAI provides survival benefit to some patients with follicular carcinoma of the thyroid. The greatest improvements were seen in those patients with locoregional or distant disease spread. (Surgery 2005;138:1072-7.) From the Departments of General Oncologic Surgery and Biostatistics, City of Hope National Medical Center, Duarte, Calif
FOLLICULAR CARCINOMA OF THE THYROID accounts for approximately 15% of the 25,690 cases of thyroid cancers annually.1 In most analyses, it is grouped with the more common papillary type of thyroid cancer. As with papillary cancer, risk stratification schemes with patient age, grade of neoplasm, presence of metastatic disease, tumor size, extent of disease, and completeness of resection often are used to determine a patient’s risk of morbidity and death from follicular carcinoma, with the purpose of helping to guide operative and medical therapies.2-4 Survival after treatment for follicular carcinoma of the thyroid, although not superior
to that of papillary cancer, is generally very good, with 20-year overall survivals exceeding essentially all other visceral malignancies. Because it is most often studied in conjunction with papillary carcinoma, less is known about follicular cancer and its biologic behaviors and response to various treatment regimens. Specifically, controversies exist regarding the optimal extent of surgical resection and the role of postoperative radioactive iodine (RAI). This study sought to determine whether the use of RAI improved overall survival in patients with follicular carcinoma of the thyroid. Additionally, patients were assessed by risk to determine which, if any, groups benefit from RAI.
Accepted for publication September 3, 2005.
MATERIAL AND METHODS The Surveillance, Epidemiology, and End Results (SEER) database is a National Cancer Institute-designed and --maintained catalog of patient data. It is a representative sample of the United States urban population, encompassing approximately 14% of the total population that resides in Atlanta, Detroit, San Francisco, Seattle, Los Angeles, San Jose, Connecticut, Hawaii, Iowa, New Mexico, and
Presented at the 26th Annual Meeting of the American Association of Endocrine Surgeons, Cancun, Mexico, April, 2005. Reprint requests: Joshua D. I. Ellenhorn, MD, Department of General Oncologic Surgery, City of Hope National Medical Center, 1500 E Duarte Rd, Duarte, CA 91010 E-mail: [email protected]. 0039-6060/$ - see front matter Ó 2005 Mosby, Inc. All rights reserved. doi:10.1016/j.surg.2005.09.021
1072 SURGERY
Podnos et al 1073
Surgery Volume 138, Number 6
Fig 1. Overall survival of all follicular patients (P = .104).
Table I. Multivariate analysis Lower Upper Risk ratio CI CI P value Metastatic disease Lymph node disease Lack of radioisotope Tumor size <2 cm Age <45 y
2.47 1.94 1.24 0.72 0.34
1.80 1.47 1.03 0.52 0.24
3.31 2.50 1.51 0.96 0.46
<.0001 <.0001 .026 .022 <.0001
Utah. The SEER database was queried for those patients with a diagnosis of follicular thyroid carcinoma with the International Classification of Diseases for Oncology, ninth revision (C73.9). Patients with histologic subtypes of follicular thyroid carcinoma coded 8330, 8331, and 8332 are included in statistical analysis. No Hu¨rthle cell or follicular variants of papillary cancer are included in the analysis. Patient data from 1973 to 2001 was searched; however, only the data for patients that were entered during or after 1988 contain the necessary information and are included for comparison. Specific data gleaned from SEER with the use of SEERStat included age, primary tumor size, the presence of lymph node metastasis, the presence of distant metastasis, the operation that was performed, the medical therapies that were used, and gender. Data regarding operative treatment, tumor size, disease spread, and specific use of RAI were available only after 1988. Information regarding the recurrence of neoplasm, the chronology of the procedures that were performed, neoplastic invasion, and RAI response were not available in the database. Statistical analyses were performed with (SAS, Cary, NC). Statistical significance was defined as a probability value of <.05. Data were plotted using (Insightful, Seattle, WA).
Fig 2. Survival of patients with distant metastases (P = .028).
RESULTS A total of 4317 patients with follicular thyroid carcinoma were identified from the SEER database from 1973 to 2001. Of these, 2112 patients (1085 patients received RAI; 1027 patients received no radiation) were in the group that were diagnosed in 1988 or later and are used for analysis. The median follow-up time for this group was 95 months. The 14-year overall survival rate for patients with follicular cancer was 80% for patients who received RAI compared with 71% for patients who did not receive RAI (P = .104). Overall survival curves are shown in Fig 1. Multivariate analysis was then performed to determine which factors improved or adversely affected overall survival. The lack of RAI was associated with a risk ratio of 1.24 (95% CI, 1.031.51; P = .026). Distant metastatic disease, the greatest predictor of poor overall survival, was associated with a risk ratio of 2.47 (95% CI 1.80-3.31; P < .0001); disease that had spread to the cervical lymph nodes was associated with a relative risk of 1.94 (95% CI, 1.47-2.50; P < .0001). These 2 factors were associated with one another (P < .0001). Factors associated with improved risk included primary tumor size <2 cm (relative risk, 0.72; 95% CI, 0.52-0.97; P = .022) and age <45 years (relative risk, 0.34; 95% CI, 0.24-0.46; P < .0001). These are summarized in Table I. For each of the aforementioned factors that were found on multivariate analysis to affect significantly overall survival at 14 years, comparisons were made regarding RAI treatment and survival. Treatment with RAI in patients with distant metastases more than doubled overall survival from 17.6% to 38.3% (P = .036; Fig 2). For patients without metastatic disease, a smaller increase in survival from 71.6% to 80.5% was seen with RAI therapy (P = .036). Of
1074 Podnos et al
Surgery December 2005
Table II. Overall survival by risk group Age <45 y Age >45 y Lymph node (ÿ) Lymph node (+) Metastases (ÿ) Metastases (+) Size <2 cm Size >2 cm
Fig 3. Overall survival of patients whose results were lymph node positive (P = .018).
the 76 patients with cervical lymph node disease, the 49 patients who received RAI had a 14-year survival of 55%, compared with 0% for the 27 patients who were treated without RAI (P < .02; Fig 3). Treatment with RAI improved survival in patients with tumor size <2 cm from 87.5% to 96.8% (P = .041). Independently, age >45 years, age <45 years, nor tumor size >2 cm were associated with improved overall survival after RAI. However, large tumor size was associated with older age (P < .05). The full results are summarized in Table II. In each analysis based on size, age, presence of cervical lymph node disease, and presence of distant metastases, several groups were found to benefit from RAI. The most dramatic survival difference occurred in patients with positive lymph node disease, distant metastases, tumors >2 cm, and age >45 years. In this group of only 11 patients, median survival in the 8 patients after RAI was 57 months, compared with only 1 month in the 3 patients who did not receive RAI (P < .01). Patients with distant metastatic disease and tumors >2 cm also benefited from RAI. In this group of 42 patients, the 10 patients who did not receive RAI had a median survival of 17 months, compared with 57 months in the 32 patients who received RAI (P < .01). In an assessment of how the extent of operation affected survival, each permutation of size < or >2 cm, age > or <45 years, the presence or absence of cervical lymph node disease, and the presence or absence of distant metastasis were examined. The only subgroup that derived a survival benefit was patients with all of the following characteristics: tumor size >2 cm, age >45 years, and no distant metastases. In that group of 49 patients, 44 patients underwent total thyroidectomy, and 5 patients underwent a less aggressive resection. Median survival
+RAI (%)
No RAI (%)
P value
95.5 63.5 81.9 54.8 80.5 38.3 96.8 76.6
93.7 53.6 72.5 0 71.6 17.6 87.5 68.3
.959 .233 .588 .018 .036 .028 .041 .213
in those patients who underwent total thyroidectomy was 57 months, compared with 32 months with less thorough resection (P < .04 by Wilcoxon test; P = .09 by log-rank). DISCUSSION Together with papillary and Hu¨rthle cell carcinoma, follicular carcinomas of the thyroid are referred to commonly as well-differentiated thyroid cancer. Follicular carcinoma accounts for nearly 15% of all thyroid cancers, although papillary and its variants comprise approximately 80% of cases. Risk stratifications systems (AMES, AGES, MACIS) were developed with the use of data sets that included both of these major histologic categories in a single group.2-4 However, various thyroid histologic conditions have distinct biologic behavior. It is known that papillary cancers most often spread through the lymphatic system,5,6 whereas follicular carcinoma has a predilection for hematogenous spread.7 Despite the fact that both papillary and follicular cancer share older age, distant metastases, cervical lymph node disease, and large tumors as poor prognostic factors, the difference in overall survival both with and without RAI underscores the different biologic composition of these tumors. In grouping the 2 diseases together, the resulting conclusions overwhelmingly under-represent follicular variants and render the conclusions difficult to extrapolate to the treatment of patients with follicular cancers. Thyroid carcinoma survival, although often excellent, differs greatly between papillary and follicular types. In a large singleinstitution review, survival for papillary cancer at 10 and 20 years was 96% and 95%, respectively.3 For follicular cancer, 10- and 20-year overall survival was 87% and 71%, respectively. Patients with follicular carcinoma in the lower risk stratifications had 20-year survival rates of 86%, although high-risk patients had only 8% 20-year survival. These patients died of metastases to the lungs, bones, brain,
Surgery Volume 138, Number 6
spinal cord, and neck. In that study, cervical lymph node disease, although uncommon, was associated with a worse prognosis, because 60% of these patients also had distant metastatic disease. Our data confirmed this finding and could explain the reasons that these patients with cervical lymph node disease endured a worse overall survival. The use of RAI in patients with well-differentiated thyroid cancer is well-established in the literature. Because of its predominance, the findings of these studies are more applicable to papillary than follicular carcinoma. Even when pooled as a meta-analysis, the benefit of RAI in low-risk patients could not be verified definitively.8 In the present study, RAI appeared to provide a survival advantage when analyzed in a multivariate model. However, the survival benefit of RAI is not uniform. Patients in higher risk groups (the presence of lymph node or distant metastases) derive a survival advantage from RAI, although patients with larger tumors or older age do not. These latter 2 traits are known risk factors in follicular cancer and often are associated with one another. The diminished benefit of RAI with increasing age may relate to the finding that, in older patients (and those with larger tumors), uptake of RAI is less predictable than in younger patients.9 This may explain the reason that, contrary to expectation, older patients and patients with larger tumors are not helped significantly by RAI. As with papillary carcinoma, the adjuvant use of RAI in patients with follicular cancers has never been proved in a prospective randomized trial. In a retrospective review of 135 patients with a median follow-up time of 10.8 years, postoperative RAI in patients with follicular cancer did not reduce the risk of distant recurrence.10 Samaan et al11 found a significant decrease in recurrence but no change in mortality rate or disease-free interval in the 108 patients in their study with follicular carcinoma. In the SEER database of 2112 patients who were entered since 1988, overall survival was improved with RAI by multivariate analysis. The survival benefit extended across some risk stratification groups. By multivariate analysis on the entire data set, these data show that the failure to use adjuvant RAI is associated with a 24% increase in disease-specific mortality rates. On univariate analysis, a trend toward statistical significance existed. Inspection of the graph seen in Fig 1 shows the curves to begin divergence at approximately 10 years. This is not surprising, given the fact that follicular cancer generally is slow growing. Patients who did not receive RAI or those for whom RAI will not result in a survival advantage are more likely die of the disease after approximately 10 years. It is likely that the curves
Podnos et al 1075
will continue to separate, rendering the graph statistically significant in time. Although substantial, it is possible that the survival benefit of RAI was missed by the analysis of previous data sets that were smaller than SEER. Because the SEER database does not include recurrence, we were unable to address how RAI affected recurrence. We can hypothesize, however, that RAI would eradicate remnant tumor nests that lead to a reduction in tumor recurrence. However, data regarding the uptake of RAI and how it affected patient treatment are not included in the SEER database. Clearly, this question is not answered definitively, and additional studies of the effect of adjuvant RAI on tumor recurrence are necessary. The extent of operation required for patients with follicular carcinoma remains somewhat controversial. Some surgeons perform total thyroidectomy on these patients with the sole purpose of being able to provide adjuvant RAI or monitor thyroglobulin postoperatively as a tumor marker.12 Other surgeons advocate total thyroidectomy because these patients with follicular cancers already are placed in the intermediate or high-risk category (along with papillary cancers that have some poor prognostic variables).13 However, a number of studies found no difference in overall survival in patients with follicular cancer by the type of operation that was performed.7,14 In the evaluation of the entire SEER database, multivariate analysis does not reveal any statistically significant amelioration of overall survival on the basis of the operation that was performed. Assessment of groups on the basis of risk factors revealed that the only patients who derived survival benefit from total thyroidectomy were those with the full constellation of older age, large tumors, and distant metastases. Although the log-rank result was not significant (P = .09), statistical significance was achieved with the Wilcoxon test (P = .04). The latter test weighs events that occur early in the analysis more than those events that occur subsequently. As a result, this analysis shows that, despite no overall significant improvement in survival, total thyroidectomy was associated with a reduction in earlier deaths. Despite the recommendation that patients with follicular carcinoma undergo total thyroidectomy, this does not appear to be practiced universally. Haigh et al15 found that patients with follicular thyroid cancer were actually less likely to undergo total thyroidectomy. Additionally, contrary to every well-known risk stratification system, older patients were found to be less likely to undergo more extensive thyroid resections. This paradoxic finding
1076 Podnos et al
might relate to controversy about the extent of operation and adjuvant role of RAI. In clarifying the utility of RAI in most patients with follicular thyroid carcinoma, our results should help alleviate some of the confusion and lack of uniformity with respect to how these patients are treated. With the many considerations that are entertained when treating patients with follicular carcinoma of the thyroid, the decision about which therapies should be used still requires attention to an individual’s risk-factor profile. REFERENCES 1. Jemal A, Murray T, Ward E, Samuels A, Tiwari RC, Ghafoor A, et al. Cancer statistics, 2005. CA Cancer J Clin 2005;55: 10-30. 2. Cady B, Sedgwick C, Meissner WA, Wool MS, Salzman FA, Werber J. Risk factor analysis in differentiated thyroid cancer. Cancer 1979;43:810-20. 3. Hay ID, Grant CS, Taylor WF, McConahey WM. Ipsilateral lobectomy versus bilateral lobar resection in papillary thyroid carcinoma: a retrospective analysis of surgical outcome using a novel prognostic scoring system. Surgery 1987;102: 1088-92. 4. Hay ID, Bergstralh EJ, Goellner JR, Ebersold JR, Grant CS. Predicting outcome in papillary thyroid carcinoma: development of a reliable prognostic scoring system in a cohort of 1779 patients surgically treated at one institution during 1940 through 1989. Surgery 1993;114:1050-8. 5. Shah JP. Differentiated thyroid cancer. In: Bloom HJG, Hanham IWF, Shaw HJ, editors. Head and neck oncology. New York: Ravens Press; 1986. p. 207-14. 6. Attie JN, Khafif RA, Steckler RM. Elective lymph node dissection in papillary carcinoma of the thyroid. Am J Surg 1971;122:464-71. 7. Brennan MD, Bergstralh EJ, van Heerden J, McConahey WM. Follicular thyroid cancer treated at the Mayo Clinic 1946 through 1970: initial manifestations, pathologic findings, therapy, and outcome. Mayo Clin Proc 1991;66:11-22. 8. Sawka AM, Thephamongkhol K, Brouwers M, Thabane L, Browman G, Gerstein HC. A systematic review and metaanalysis of the effectiveness of radioactive iodine remnant ablation for well-differentiated thyroid cancer. J Clin Endocrinol Metab 2004;89:3668-76. 9. Apostoaei AI, Miller LF. Uncertainties in dose coefficients from ingestion of 131I, 137Cs, and 90Sr. Health Phys 2004; 86:460-82. 10. Chow SM, Law SC, Mendenhall WM, Au SK, Yau S, Yuen KT, et al. Follicular thyroid carcinoma: prognostic factors and the role of radioiodine. Cancer 2002;95:488-98. 11. Samaan NA, Maheshwari YK, Nader S, Hill CS Jr, Schultz PN, Haynie TP, et al. Impact of therapy for differentiated carcinoma of the thyroid: an analysis of 706 cases. J Clin Endocrinol Metab 1983;56:1131-8. 12. Duren M, Siperstein AE, Shen W, Duh QY, Morita E, Clark OH. Value of stimulated serum thyroglobulin levels for detecting persistent or recurrent differentiated thyroid cancer in high- and low-risk patients. Surgery 1999;126:13-9. 13. Shaha AR, Loree TR, Shah JP. Intermediate-risk group for differentiated carcinoma of thyroid. Surgery 1994;116: 1036-41. 14. Wanebo H, Coburn M, Teates D, Cole B. Total thyroidectomy does not enhance disease control or survival even in high-
Surgery December 2005
risk patients with differentiated thyroid cancer. Ann Surg 1998;227:912-21. 15. Haigh PI, Urbach DR, Rotstein LE. AMES prognostic index and extent of thyroidectomy for well-differentiated thyroid cancer in the United States. Surgery 2004;136:609-16.
DISCUSSION Dr Subhash Patel (Chicago, Ill). This is a very interesting paper. However, it is limited by the retrospective nature, especially when it comes to lymph node status. Routine lymph node dissection is still controversial and is not performed by most surgeons. Therefore, any analysis or conclusions that are drawn from incomplete data to determine the importance of RAI therapy in relation to lymph node status would be erroneous. Dr Podnos. The problem with thyroid cancer is that we will never get any prospective randomized studies to really answer these questions well. Because of the long nature of the course of the disease, it makes any prospective studies unfeasible. So we must use databases like SEER and single-institutional studies the best that we can. Dr Patel. The lymph node status remains unclear. Lymph node dissection probably was performed in selected patients in whom either they were clinically palpable or appeared large at the time of surgery. What about those patients in whom the lymph nodes are not removed but who may have microscopic metastasis? The survival benefit of RAI therapy in these patients could not be analyzed in this retrospective review. Dr Podnos. That is a good question. Because this is a large database, I cannot account for what happened to each of these patients. It is likely that those patients who were clinically positive regarding lymph node disease but only microscopically positive were in the database as being negative, and nothing has been done about them. That is an interesting point, because that is where the use of RAI in theory could come to help. In papillary cancer, we presented a paper at the SSO that showed that RAI really did not help very much. In follicular cancer, it might help some people and not others. Dr Peter Angelos (Chicago, Ill). In your analysis of the data, can you assess the survival advantage to the use of RAI versus the extent of the operation? In other words, are you comparing total thyroidectomy with RAI to total thyroidectomy without RAI or are you comparing total thyroidectomy with RAI to lobectomy? Dr Podnos. We ran this analysis also based on the operation that was performed, and the operation that was performed did not make a difference, except in one very small, specific group of patients who I did not present here but who are in the article. Those patients underwent a total thyroidectomy and no RAI, and there was no difference. We think that these results are most likely a function of the RAI rather than the operation that was performed. Dr Keith S. Heller (Lake Success, NY). The data are the data, and you have to analyze what you have. Are you concerned that perhaps the SEER database, because of the way it is collected, underreports the use of a treatment such as RAI? I ask because RAI frequently is not
Surgery Volume 138, Number 6
given in the hospital, and most of the SEER data comes from hospital records. Looking at your numbers, specifically for patients with distant metastases, it is difficult for me to believe that there are that many patients with metastatic follicular cancer who did not receive RAI. I suspect it is not picked up in the database. If that group is underreporting the use of RAI, I would be concerned that all your other groups are underreporting the use of RAI as well. Dr Podnos. That, too, was a big concern. That is the reason that, with the last slide, I tried to mention some of the problems with SEER. There definitely are some problems with it, and they need to be understood when looking at the data. There could be decreases in reporting of RAI. In looking at those graphs, if the groups were really the same, I would expect there not to be any differences. The differences in patients with metastatic disease were really quite dramatic. I would tend to believe them. Dr Ashok R. Shaha (New York, NY). It is difficult to look at the data from SEER and not be critical about it. The numbers are large. There has been a flurry of papers in the recent literature from SEER because this is a public domain. You can go into the website and get all the information and try to analyze, reanalyze, and reanalyze. I think we must be very cautious when we make important conclusions based on these data. One of the important factors in follicular thyroid cancer is capsular invasion; that information is not
Podnos et al 1077
available. Anytime we interpret the data based on what we have, we may be drawing some incorrect conclusions. This is a good demographic paper, but I think drawing conclusions would be very difficult. Based on your last slide, I think you may want to conclude that RAI offers improvement in patients with high-risk follicular carcinoma of the thyroid, which that may be easier to conclude than the title as it stands right now. Dr Podnos. As I said, I agree that there are some problems with SEER and that these must be understood in interpreting the data. In terms of capsular invasion, that is something that has been well-established as being a risk factor. It is not in SEER. There are other things that we mentioned that were not overlooked that are in SEER and must be confirmed in subsequent papers. For that type of thing, SEER is very beneficial. Dr Henning Dralle (Halle, Germany). I am a little concerned about the follow-up. I have a very short question regarding the new UICC classification. Did you observe any difference in the group <2 cm with regard to survival or recurrence rates? Dr Podnos. We cannot speak to recurrence because that was not in SEER. We broke it down by size > or <2 cm because that is what TNM states. When size was run as a continuous variable, the results were the same. Dr Dralle. But there are T categories 1-A and 1-B, so you can perhaps analyze these groups as well.