Evaluation of Overall Survival in Patients With Anaplastic Thyroid Carcinoma, 2000-2019

Key Points

Question  Is emerging use of targeted therapy, immunotherapy, surgery, and radiation therapy associated with improved overall survival in patients with anaplastic thyroid carcinoma (ATC)?

Findings  In this single-institution cohort study of 479 patients with ATC spanning nearly 20 years, 1- and 2-year survival significantly increased from 35% and 18% in the 2000-2013 era (n = 227) to 47% and 25% in the 2014-2016 era (n = 100), and 59% and 42% in the 2017-2019 era (n = 152), respectively.

Meaning  This study suggests that ATC may be effectively treated with highly specialized molecular-based personalized therapies, and surgery when appropriate, regardless of disease stage.

Importance  Anaplastic thyroid carcinoma (ATC) historically has a 4-month median overall survival (OS) from time of diagnosis, with disease-specific mortality approaching 100%. The association between recent major advancements in treatment and OS has yet to be evaluated.

Objective  To evaluate rates of OS in patients with ATC over the last 2 decades.

Design, Setting, and Participants  Retrospective cohort study in a single tertiary care institution. Patients with histopathological confirmation of ATC from January 2000 to October 2019 were included and divided into 3 groups according to date of presentation: 2000-2013, 2014-2016, and 2017-2019.

Main Outcomes and Measures  Overall survival compared among different treatment eras and differing therapies, including targeted therapy, immunotherapy, and surgery.

Results  Of 479 patients (246 men [51%]; median age, 65.0 [range, 21.1-92.6] years) with ATC evaluated, 52 (11%) were stage IVA, 172 (36%) stage IVB, and 255 (53%) stage IVC at presentation. The median OS of the entire cohort was 0.79 years (9.5 months), ranging from 0.01 to 16.63. The OS at 1 and 2 years was 35% (95% CI, 29%-42%) and 18% (95% CI, 13%-23%) in the 2000-2013 group (n = 227), 47% (95% CI, 36%-56%) and 25% (95% CI, 17%-34%) in the 2014-2016 group (n = 100), and 59% (95% CI, 49%-67%) and 42% (95% CI, 30%-53%) in the 2017-2019 group (n = 152), respectively (P < .001). The hazard ratio was 0.50 (95% CI, 0.38-0.67) for the 2017-2019 group compared with the 2000-2013 patients (P < .001). Factors associated with improved OS included targeted therapy (hazard ratio, 0.49; 95% CI, 0.39-0.63; P < .001), the addition of immunotherapy to targeted therapy (hazard ratio, 0.58; 95% CI, 0.36-0.94; P = .03), and surgery following neoadjuvant BRAF-directed therapy (hazard ratio, 0.29; 95% CI, 0.10-0.78; P = .02). Patients undergoing surgery following neoadjuvant BRAF-directed therapy (n = 20) had a 94% 1-year survival with a median follow-up of 1.21 years.

Conclusion and Relevance  In this large single-institution cohort study spanning nearly 20 years, changes in patient management appear to be associated with significant increase in survival. The era of untreatable ATC is progressively being replaced by molecular-based personalized therapies, with integration of multidisciplinary therapies including surgery and radiation therapy.

Anaplastic thyroid carcinoma (ATC) remains one of the most aggressive and fatal solid tumors. The most recent Surveillance, Epidemiology, and End Results database analysis (1986-2015) reported median overall survival (OS) of 4 months and 6-month OS of 35%,¹ while disease-specific mortality is 98% to 99%.^2,3 Unlike differentiated thyroid cancer, which has a good prognosis and is managed primarily with surgery, patients with ATC usually present with a rapidly growing and invasive neck mass, regional cervical lymph node involvement, while approximately half of the patients will also have distant metastatic disease,^4,5 at presentation. Consequently, patients with ATC are usually inoperable at presentation and have historically been treated palliatively or referred to hospice. As such, although ATC comprises less than 2% of thyroid cancers, it is estimated to be the cause of more than 50% of annual thyroid cancer-related mortality.⁴

In 2014, the Facilitating Anaplastic Thyroid Cancer Specialized Treatment (FAST) team⁶ was developed at The University of Texas MD Anderson Cancer Center, which provided patients with rapid access to multidisciplinary care and tumor molecular testing. Subsequently, many patients were enrolled in therapeutic clinical trials, which demonstrated substantial survival improvement with combination dabrafenib and trametinib therapy in patients harboring the BRAFV 600E variation. This clinical trial ultimately led to the first US Food and Drug Administration (FDA)–approved drug combination for ATC.⁷ Figure 1 summarizes the current ATC FAST program treatment algorithm at The University of Texas MD Anderson Cancer Center.

Because a subset of patients achieved significant response to targeted therapy, they became candidates for additional multimodal adjunct therapies, such as surgery^8,9 and intensity-modulated radiation therapy, therapies that historically had little impact on survival. Recent work has demonstrated feasibility of complete resection and locoregional disease control when patients with BRAF V600E-variant tumors undergo surgical resection following neoadjuvant BRAF-directed therapy.⁸

With accelerated access to highly specialized care, improved understanding and testing of tumor genetics, emergence of novel targeted therapies and immunotherapies—particularly in the setting of clinical trials—and the integration of surgical resection in ATC patient management, specifically after neoadjuvant therapy, we hypothesized significant improvement in OS of patients with ATC over the past 2 decades, and especially over the past 5 years.

We retrospectively evaluated consecutive patients with pathologically confirmed diagnosis of ATC who presented to The University of Texas MD Anderson Cancer Center from January 2000 to October 2019. Data were obtained from an institutional database following institutional review board approval. All patients had histopathologically confirmed diagnosis of as ATC. Patient demographic characteristics, age-adjusted Charlson comorbidity index,¹⁰ and data on BRAF V600E molecular testing and treatment modalities (cytotoxic chemotherapy, targeted therapy, immunotherapy, surgery, and radiation therapy) were collected. The OS was measured from date of pathologic confirmation of ATC to date of death, with patients censored at date of last follow-up.

Patients were divided into 3 subgroups according to date of presentation to The University of Texas MD Anderson Cancer Center: January 2000 to December 2013, January 2014 to December 2016, and January 2017 to October 2019. The choice of year to separate the groups corresponds to changes in our practice: 2014 marked initiation of the FAST program,⁶ and 2017 marked the beginning of the integration of surgery and radiation performed after neoadjuvant therapies at The University of Texas MD Anderson Cancer Center, as well as the first full year of a multimodal targeted therapy plus immunotherapy trial (NCT03181100).¹¹

Two-tailed Pearson χ² test (or Fisher exact test when categorical variable values were ≤5) was used to analyze categorical variables. A 1-way analysis of variance was used to analyze continuous variables. The Kaplan-Meier method was used to estimate OS, while log-rank test was used to assess between-group differences in OS. A Cox proportional hazards regression model and the Efron method of tie handling were used to assess relative risk (hazard) of death, adjusting for age, sex, comorbidity, and disease stage. Cox proportional hazards regression model assumption tests were based on Schoenfeld residuals. All statistical analyses were performed using Stata, version 13.0 (StataCorp LP).

From January 1, 2000, 479 consecutive new patients (233 women [49%], 246 men [51%]; median age, 65.0 [range, 21.1-92.6] years) with histopathological confirmation of ATC presented to The University of Texas MD Anderson Cancer Center; 52 (11%) were stage IVA, 172 (36%) stage IVB, and 255 (53%) stage IVC at presentation. The 2000-2013 group comprised 227 patients, the 2014-2016 group 100 patients, and the 2017-2019 group 152 patients. There was no statistically significant difference between the 3 groups in terms of age, sex, comorbidity, and disease stage (Table 1).

The median number of new patients with ATC evaluated per year was 23 (range, 9-55) (eFigure 1 in the Supplement), excluding 2019, wherein 44 patients were seen up to October 2, 2019. The mean number of new patients with ATC seen annually from 2000 to 2019 increased by 212% when comparing the 2000-2013 group vs the 2017-2019 group.

In terms of molecular testing, 268 (56%) patients were tested for the BRAF V600E variation either through immunohistochemistry or next-generation sequencing on tissue or blood (cell-free DNA). Specifically, 17% (38 of 227) of the 2000-2013 group, 82% (82 of 100) of the 2014-2016 group, and 97% (148 of 152) of the 2017-2019 group patients had BRAF testing, (P < .001). Prevalence of ATC tumors with the BRAF V600E variation among tested patients was not significantly different between the 3 groups, with an overall rate of 38% (32% in the 2000-2013 group, 41% in the 2014-2016 group, and 37% in the 2017-2019 group) (P = .57).

Two patients were excluded from survival analyses, 1 without follow-up, and another who presented with a remote history of ATC dating to 1987; therefore, 477 patients were included in the survival analyses. The median OS was 0.79 years (9.5 months), ranging from 0.01 to 16.63 years. Six-month, 1-year, and 2-year OS for the entire cohort were 67% (95% CI, 63%-71%), 44% (95% CI, 40%-49%), and 25% (95% CI, 21%-29%), respectively.

The OS at 1 year was 35% (95% CI, 29%-42%) in the 2000-2013 group, 47% (95% CI, 36%-56%) in the 2014-2016 group, and 59% (95% CI, 49%-67%) in the 2017-2019 group. This finding translates into a 1-year survival improvement of 12% (95% CI, 0%-24%) in the 2014-2016 group and 24% (95% CI, 13%-35%) in the 2017-2019 group. The OS estimates at 2 years were 18% (95% CI, 13%-23%), 25% (95% CI, 17%-34%), and 42% (95% CI, 30%-53%), respectively (P < .001) (Figure 2), with adjusted hazard ratios of 0.77 (95% CI, 0.59-1.01; P = .06) when comparing the 2000-2013 group with the 2014-2016 group and 0.50 (95% CI, 0.38-0.67; P < .001) when comparing the 2000-2013 group with the 2017-2019 group (Table 2). This translates into a 2-year survival improvement of 7% (95% CI, −2% to 16%) in the 2014-2016 group and 24% (95% CI, 13%-35%) in the 2017-2019 group. The Cox proportional hazards regression assumption was met (P = .46). The median OS according to group was 0.67 years (8.0 months) in the 2000-2013 group, 0.88 years (10.6 months) in the 2014-2016 group, and 1.31 years (15.7 months) in the 2017-2019 group. Median group-specific follow-up time from diagnosis was 0.61 years (range, 0.01-16.63 years) in the 2000-2013 group, 0.74 years (range, 0.02-5.68 years) in the 2014-2016 group, and 0.64 years (range, 0.01-5.16 years) in the 2017-2019 group. The total number of deaths was 197 in the 2000-2013 group, 79 in the 2014-2016 group, and 62 in the 2017-2019 group.

When separating patients according to clinical stage at presentation, OS rates at 2 years according to the 3 eras (2000-2013, 2014-2016, and 2017-2019) were 36% (95% CI, 19%-54%), 78% (95% CI, 16%-97%), and 63% (95% CI, 29%-85%) for stage IVA; 24% (95% CI, 15%-34%), 27% (95% CI, 13%-42%), and 44% (95% CI, 26%-61%) for stage IVB; and 9% (95% CI, 4%-15%), 18% (95% CI, 9%-30%), and 37% (95% CI, 22%-52%) for stage IVC, respectively (P < .001) (eFigure 2 in the Supplement).

Targeted therapy included dabrafenib, trametinib, vemurafenib, cobimetinib, larotrectinib, everolimus, pazopanib, bevacizumab, lenvatinib, selpercatinib, lenalidomide, and cetuximab. Immunotherapy drugs were usually used in combination with targeted therapy or chemotherapy and included pembrolizumab, atezolizumab, nivolumab, and ipilimumab. Significantly more patients received targeted therapy in the 2017-2019 group vs the 2000-2013 and 2014-2016 groups (61%, 9%, and 43%, respectively; P < .001). When including only patients having been tested for the BRAF V600E variation, these values were 65%, 26%, and 46%, respectively (P < .001). Median OS for patients treated with targeted therapy, regardless of their grouping, was 1.31 years (15.7 months) (95% CI, 1.07-1.99 years) compared with 0.63 years (7.6 months) (95% CI, 0.52-0.72 years) in patients not having received any targeted therapy, with an adjusted hazard ratio of 0.49 (95% CI, 0.39-0.63; P < .001) (eFigure 3 in the Supplement), which translates into a survival improvement of 0.68 years (8.2 months) (95% CI, 0.33-1.03 years). The use of immunotherapy also increased significantly in the 2017-2019 group (47%) compared with the 2000-2013 (1%) and 2014-2016 (18%) groups (P < .001). For patients seen between 2014 and 2019, those having received targeted therapy with immunotherapy (n = 76) had significantly higher OS vs those having received targeted therapy without immunotherapy (n = 59) (1.99 years [23.9 months], 95% CI 1.23-3.93 years vs 1.20 years [14.4 months], 95% CI 0.70-1.70 years), with a hazard ratio of 0.58 (95% CI, 0.36-0.94; P = .03), which translates into a survival improvement of 0.79 years (9.5 months) (95% CI, −0.82 to 2.40). There was no statistically significant difference in the proportion of patients receiving cytotoxic chemotherapy, locoregional radiation therapy, and radiation therapy to other sites between the 3 groups.

Neoadjuvant therapy included atezolizumab/cobimetinib (n = 2), docetaxel/doxorubicin (n = 1), and BRAF-directed therapies (combination BRAF plus MEK inhibitor): dabrafenib/trametinib (n = 13) or vemurafenib/cobimetinib (n = 7). Surgical resection following neoadjuvant therapy was completed in 23 patients (15%) in the 2017-2019 group compared with none in the 2000-2013 and 2014-2016 groups (P < .001). In patients who received BRAF-directed therapy, those who underwent surgery following neoadjuvant therapy (n = 20) had statistically significant improvement in median OS (median OS, not reached; 95% CI, 1.43-N/A years) compared with those who did not (n = 35) (median OS, 0.80 years; 95% CI, 0.67-0.86 years). The hazard ratio was 0.29 (95% CI, 0.10-0.78) in patients who underwent neoadjuvant BRAF-directed therapy followed by surgery (P = .02) (Figure 3), while the estimated 1-year OS was 94% and survival improvement at 1 year was 45% (95% CI, 18%-72%). In 12 patients, surgery included total thyroidectomy and central compartment dissection with bilateral (n = 8) or unilateral (n = 4) lateral neck dissection. In 4 patients, surgery included lobectomy alone (n = 1) or with central compartment dissection (n = 2) and with unilateral lateral neck dissection (n = 1). Four other patients had a history of surgery before receiving BRAF-directed therapy and underwent revision central neck dissection with unilateral (n = 3) or bilateral (n = 1) lateral neck dissection. Lateral neck dissections included levels 2, 3, 4, and 5B, and overall 17 patients (85%) had either unilateral or bilateral neck dissection. Eleven patients (55%) received postoperative external beam radiation therapy to the thyroid bed as well as to the bilateral (n = 6) or unilateral (n = 4) lateral neck or mediastinum (n = 1). Sixteen of the 20 patients (80%) who underwent surgery following neoadjuvant BRAF-directed therapy were alive at last follow-up, with median follow-up from diagnosis of 1.21 years (range 0.26-2.70 years). Eight of 20 (40%) presented with stage IVC disease.

This cohort study of 479 patients represents, to our knowledge, the largest single-institution cohort of consecutive patients with ATC in the published literature and demonstrates significant improvement in survival over the last 2 decades. Survival at 2 years increased from 18% in the 2000-2013 group to 25% in the 2014-2016 group and 42% in the 2017-2019 group (hazard ratio, 0.50), and more than half of these patients presented with stage IVC disease. Further, since the beginning of the FAST program,⁶ there has been a significant increase in new ATC patient referrals. As ATC incidence rates have been reported unchanged over the last 30 years,¹ this increase in ATC presentations to a single institution may be owing to referral bias rather than a true increase in incidence.

A notable shift during this time was the improved understanding of distinct molecular classifications of ATC, with immediate molecular testing at patient presentation becoming standard of care in the 2014-2019 era, and nearly all patients undergoing testing in the 2017-2019 era. Knowledge of oncogenic molecular factors was used to guide systemic therapy, particularly the use of targeted therapy and neoadjuvant BRAF-directed therapy before surgical resection. The use of targeted therapy increased significantly over time, which was associated with improved median OS (1.31 years in those receiving targeted therapy vs 0.63 years in those who did not, hazard ratio, 0.49). Neoadjuvant BRAF-directed therapy before surgery (which was performed only in the 2017-2019 era) also was associated with improved survival. Comparing patients who received BRAF-directed therapy with or without subsequent surgery, the median OS was not reached in the former cohort (median follow-up 1.21 years), while median OS was 0.80 years in the latter cohort (hazard ratio, 0.29).

Recently, several major centers have reported experiences with select ATC treatment regimens over several decades, particularly focusing on patients who were able to undergo surgery and conventional chemoradiation therapy. The Mayo Clinic published outcomes on a selected 48 patients with ATC treated with multimodal therapy, including surgery, taxane-based chemotherapy, and radiation therapy, over a 13-year period (2003-2015), reporting a median OS of 9 months and 1-year OS of 42%.¹² The Portuguese Oncology Institute¹³ reported a mean survival of 2 months among 79 patients with ATC over 18 years (2000 to 2018), in which patients were treated with single or multimodal therapy that included surgery, radiation therapy, and chemotherapy. Using a similar treatment approach, Duke University¹⁴ reported a mean survival of 4 months and 1-year survival of 18% among 28 patients over 25 years (1990 to 2015). Memorial Sloan Kettering Cancer Center¹⁵ recently published retrospective data on a selected group of 104 patients with ATC undergoing radiation therapy over 33 years (1984 to 2017) treated either with concurrent chemoradiation therapy (doxorubicin or paclitaxel-based) or surgery and concurrent chemoradiation therapy. Median survival for the overall Memorial Sloan Kettering Cancer Center cohort was 7 months, with 1-year survival of 34%, while patients treated selectively with surgery and postoperative concurrent chemoradiation therapy (n = 53) had a 1-year survival of 55%. In our study, we report a 1-year OS of 94% among a highly selected group of 20 patients with BRAF variation undergoing neoadjuvant BRAF-directed targeted therapy followed by surgery (with or without postoperative radiation therapy) from 2017 to 2019. This is, to our knowledge, the highest OS in the reported literature among a selected cohort of newly diagnosed ATC patients, almost half of which presented with stage IVC disease.

Prioritizing clinical trial enrollment has been a key factor in advancing care for patients with ATC. While numerous national ATC-directed trials have either never reached accrual or been unable to commence enrollment, this study demonstrates clinical trial enrollment rates that have reached 34% for the most recently diagnosed ATC patients⁶—well above the national mean of less than 3% among all patients with cancer.¹⁶ Such clinical trials led to the FDA approval of dabrafenib with trametinib in 2018 for patients with BRAF V600E-variant ATC, representing the first drug therapy approved by the FDA for ATC. In addition to the BRAF V600E variation (38% positivity in this study), other clinical trials have recruited or are currently recruiting ATC patients based on the presence of genetic abnormalities such as RAS or NF1 variation (cobimetinib and atezolizumab combination [NCT03181100]),¹¹ NTRK gene fusion (larotrectinib [NCT02122913]),¹⁷ RET variation (pralsetinib [NCT03037385]; selpercatinib [NCT03157128]; BOS172738 [NCT03780517]¹⁸), as well as several other targeted and immunotherapy-based¹⁹ therapies summarized in a recent review.²⁰ In the current study, we report a significant increase in the use of immunotherapy in recent years, although survival improvements cannot be directly attributed to immunotherapy alone because of multiple confounding factors, particularly the use of concurrent targeted therapy. While single-agent immunotherapy has not shown great promise in the treatment of ATC,²¹ ongoing and/or completed clinical trials are studying the potential of combining immunotherapy agents such as pembrolizumab¹⁹ and atezolizumab¹¹ with targeted therapy, with notable results to date. The rationale for adding immunotherapy is that all patients eventually develop resistance to kinase inhibitor targeted therapy. Immunotherapy potentially delays and/or prevents the emergence of resistance variants and also affords the ability to hold targeted therapy when doing so is required (ie, due to toxicity, during surgery, during radiation), allowing for maintenance of disease control.

This study has limitations, including its retrospective design and the possibility that referral bias to a tertiary cancer center may potentially lead to higher than expected survival. Nevertheless, this study represents to date the largest single-institution experience with ATC in the literature, with more than 50% of patients presenting with stage IVC disease. Although several subgroup analyses are reported herein, the overall group of 479 patients evaluated over a 20-year period represents all patients with ATC (including >40 patients per year since 2016), while studies reporting selected subgroups of patients (with very few patients per year) must be evaluated in the context of treatment selection bias. In addition, novel targeted therapies and immunotherapy are expensive treatments, and while some of these therapies have been FDA-approved in the US, cost of treatment is a potential limitation of global widespread implementation of novel therapies for ATC.

Recent decades have seen a shift in the median OS of patients with ATC, regardless of disease stage, with the present study demonstrating nearly twice the rate of OS compared with recently published single-institution data,^12-15 as well as a 300% survival increase compared with the most recent Surveillance, Epidemiology, and End Results data analysis.¹ Recent changes in management of patients with ATC may be factors in this improvement in OS. These changes include the initiation of the FAST program,⁶ routine and rapid molecular testing for actionable oncogenic variants and fusions, increases in the availability of and enrollment in clinical trials, and the use of targeted therapy. Further, during the 2017-1019 era, the shift of implementing neoadjuvant BRAF-directed therapy followed by surgery may have led to greater improvements in OS for patients with the BRAF V600E variation. Given that BRAF-directed therapy is highly effective and that surgically unresectable tumors in patients become resectable after receipt of these drugs (although resistance may develop), the traditional trimodal therapy of surgery and adjuvant chemoradiation may be replaced in the future at other institutions with up-front BRAF/MEK inhibitors followed by surgery with or without adjuvant chemoradiation in patients with stage IVB and IVC BRAF V600E variation.

Over the past 5 years, treatment of ATC has evolved from primarily palliation and hospice care to effective highly specialized molecular-based personalized therapies and surgery when appropriate, regardless of disease stage.

Accepted for Publication: June 10, 2020.

Corresponding Author: Maria E. Cabanillas, MD, Department of Endocrine Neoplasia and Hormonal Disorders, The University of Texas MD Anderson Cancer Center, 1400 Pressler Rd, Unit 1461, Houston, TX 77030 (mcabani@mdanderson.org).

Published Online: August 6, 2020. doi:10.1001/jamaoncol.2020.3362

Author Contributions: Drs Cabanillas and Zafereo had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Maniakas and Dadu are co-first authors and contributed equally. Drs Cabanillas and Zafereo are co-senior authors.

Concept and design: Maniakas, Dadu, Busaidy, Wang, Ferrarotto, Gunn, Hofmann, Cabanillas, Zafereo.

Acquisition, analysis, or interpretation of data: Maniakas, Dadu, Busaidy, Ferrarotto, Lu, Williams, Gunn, Cote, Sperling, Gross, Sturgis, Goepfert, Lai, Cabanillas, Zafereo.

Drafting of the manuscript: Maniakas, Cote, Cabanillas, Zafereo.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Maniakas, Wang, Sperling.

Administrative, technical, or material support: Maniakas, Wang, Ferrarotto, Gunn, Gross, Zafereo.

Supervision: Maniakas, Dadu, Busaidy, Wang, Ferrarotto, Gunn, Hofmann, Cote, Sturgis, Cabanillas, Zafereo.

Conflict of Interest Disclosures: Dr Dadu reported receiving grants from Merck, Exelixis Eisai, and AstraZeneca, and receiving personal fees from Bayer outside the submitted work. Dr Ferrarotto reported receiving personal fees from Regeneron-Sanofi, Ayala Pharma, and Klus Pharma outside the submitted work. Dr Gross reported receiving personal fees from Shattuck Labs, PDS Biotechnology, Genzyme, and Intuitive Surgical, receiving grants from Regeneron outside the submitted work. Dr Lai reported receiving personal fees from Cardinal Health outside the submitted work. Dr Cabanillas reported receiving grants from Genentech, personal fees from LOXO, Ignyta, grants from Eisai and Merck outside the submitted work. No other disclosures were reported.

Meeting Presentation: This paper was presented at the 89th Annual Meeting of the American Thyroid Association; November 2, 2019; Chicago, Illinois.

Additional Contributions: We thank Maitrayee Goswami, MSc, Li Xu, PhD, and Dongmin Wei, MD, research assistants at The University of Texas MD Anderson Cancer Center, for their effort and time in collecting the data. No compensation was received outside of usual salary.