低分化甲状腺癌患者术后预后列线图模型的建立与评价

Development and evaluation of a postoperative prognostic nomogram model for patients with poorly differentiated thyroid carcinoma

摘要

背景与目的 低分化甲状腺癌（PDTC）是一种临床上较为少见但具有高度侵袭性的甲状腺恶性肿瘤类型，其生物学行为介于分化型与未分化型甲状腺癌之间，常表现为早期转移、复发率高和生存率低。目前，针对PDTC患者预后的评估主要依赖TNM分期等传统指标，尚缺乏系统化、多因素整合的个体化预测工具。列线图模型作为一种可视化、量化的预测方法，已广泛应用于多种肿瘤的预后研究，但在PDTC中的应用尚不充分。本研究旨在识别影响PDTC患者预后的关键危险因素，并基于多因素分析构建风险列线图模型，以期为临床精准判断PDTC患者术后预后提供辅助决策支持。 方法 纳入2015年1月─2020年12月在南阳医学高等专科学校第一附属医院接受手术治疗的55例PDTC患者，随访期为3年。根据随访期间肿瘤复发、转移及死亡情况，将患者分为预后良好组与预后不良组。通过单因素分析筛选与预后相关的临床特征，进一步采用多因素Logistic回归分析确定独立危险因素。基于多因素分析结果，使用R 3.5.3软件构建列线图风险预测模型，并通过ROC曲线和Bootstrap方法评估模型的预测性能和校准能力。 结果 55例患者中，随访期间有15例出现肿瘤进展，1例死亡，预后不良率为29.1%。单因素分析显示，肿瘤直径、TNM分期、局部侵犯、手术方式、脉管侵犯及神经受累在两组间差异具有统计学意义（均P<0.05）。多因素Logistic回归分析表明，肿瘤直径≥3 cm、晚期TNM分期、局部侵犯、次全切术、存在脉管侵犯和神经受累是PDTC患者预后不良的独立危险因素（均P<0.05）。以此为基础构建的列线图模型C-指数为0.794（95% CI=0.725~0.846），ROC曲线下面积为0.817，敏感度为82.26%，特异度为86.35%，模型表现出良好的区分度和准确率。 结论 肿瘤直径、TNM分期、局部侵犯、手术方式、脉管侵犯和神经受累是影响PDTC患者术后预后的重要因素。基于上述变量构建的列线图模型具有较好的预测能力，能够为PDTC的个体化风险评估和治疗策略制定提供有利参考。

Abstract

Background and Aims Poorly differentiated thyroid carcinoma (PDTC) is a relatively rare but highly aggressive type of thyroid malignancy. Its biological behavior lies between differentiated and undifferentiated thyroid carcinoma, and it is often characterized by early metastasis, high recurrence rates, and poor survival outcomes. At present, prognostic assessment for PDTC patients primarily relies on traditional indicators such as TNM staging, and there remains a lack of systematic, multi-factorial, and individualized predictive tools. As a visual and quantitative method, the nomogram model has been widely applied in the prognostic evaluation of various tumors; however, its use in PDTC remains limited. This study aims to identify key risk factors associated with poor prognosis in PDTC patients and to construct a risk prediction nomogram model based on multivariate analysis, in order to provide clinical support for individualized postoperative prognostic assessment. Methods A total of 55 PDTC patients who underwent surgical treatment at our hospital from January 2015 to December 2020 were retrospectively enrolled and followed up for three years. Based on tumor recurrence, metastasis, and mortality during the follow-up period, patients were divided into a good prognosis group and a poor prognosis group. Univariate analysis was performed to screen for clinical features associated with prognosis, followed by multivariate logistic regression to identify independent risk factors. A nomogram risk prediction model was constructed using R software (version 3.5.3), and its predictive performance and calibration were evaluated by receiver operating characteristic (ROC) curve and Bootstrap validation. Results During the 3-year follow-up, 15 patients experienced tumor progression and 1 patient died, resulting in a poor prognosis rate of 29.1%. Univariate analysis showed statistically significant differences in tumor diameter, TNM stage, local invasion, surgical approach, vascular invasion, and nerve involvement between the two groups (all P<0.05). Multivariate logistic regression identified tumor diameter ≥3 cm, advanced TNM stage, local invasion, subtotal thyroidectomy, vascular invasion, and nerve involvement as independent risk factors for poor prognosis (all P<0.05). The nomogram model constructed based on these variables demonstrated a C-index of 0.794 (95% CI=0.725-0.846), an AUC of 0.817, sensitivity of 82.26%, and specificity of 86.35%, indicating good discriminatory ability and predictive accuracy. Conclusion Tumor diameter, TNM stage, local invasion, surgical approach, vascular invasion, and nerve involvement are important factors influencing postoperative prognosis in PDTC patients. The nomogram model based on these variables exhibits strong predictive performance and may serve as a valuable tool for individualized risk assessment and therapeutic decision-making in clinical practice.

Graphical abstract

关键词

甲状腺肿瘤 / 预后 / 列线图模型

Key words

Thyroid Neoplasms / Prognosis / Nomogram model

甲状腺癌是一种常见的恶性肿瘤，主要分为分化型甲状腺癌和未分化型甲状腺癌^[1]。低分化甲状腺癌（poorly differentiated thyroid carcinoma，PDTC）是近年来新提出的一种甲状腺癌病理类型，其形态学表现及生物学特征介于分化型甲状腺癌和未分化型甲状腺癌之间，有学者猜测其可能代表分化型甲状腺癌向未分化型甲状腺癌转变的中间过程^[2]。目前关于PDTC的明确定义及分类标准尚有一定争议，且一些医师对于该病的临床特征缺乏足够的认知，导致其流行病学数据差异较大^[3]。PDTC的治疗多倾向于进行传统的甲状腺全切术加颈部淋巴结清扫，但相关治疗效果仍具有一定争议^[4]。由于诊疗方案尚不具有统一标准，且PDTC患者就诊时多为晚期，因此接受治疗后的预后结局不甚理想^[5]。大多研究中PDTC患者治疗后复发率较高，10年内生存率在50%左右^[6]。PDTC在甲状腺恶性肿瘤不良预后结局中占有极大的比重，但由于PDTC相较于其他类型的甲状腺恶性肿瘤罕见，因此该病预后相关影响因素的研究较为少见^[7]。目前国内鲜有关于PDTC的研究，鉴于此，本研究通过分析影响PDTC患者预后的因素，并以此构建列线图模型对预后结局进行预测，以期为优化临床PDTC的诊疗策略、改善患者预后结局制定提供参考。

1 资料与方法

1.1 一般资料

选取2015年1月─2020年12月南阳医学高等专科学校第一附属医院收治的55例PDTC患者为研究对象。纳入标准：(1) 符合PDTC相关诊断标准^[8]，且术后病理检查结果为PDTC；(2) 均接受根治性手术治疗，生命体征平稳；(3) 一般资料完整；(4) 术前无接受放化疗等其他治疗；(5) 依从性高，能坚持配合随访。排除标准：(1) 合并其他类型的恶性肿瘤者；(2) 神志不清，无法正常沟通者；(3) 随访期间因甲状腺癌之外的其他原因死亡者；(4) 合并血液系统、免疫系统疾病者。本研究获得本院伦理委员会批准（伦理批号：IRB-Y-2024027）。

1.2 资料收集及分组

对患者进行为期3年的随访，方式为门诊随访和电话随访，第1年每3个月随访1次，之后每6个月随访1次，将患者术后肿瘤进展（复发、转移）及死亡归为预后不良组，其余归为预后良好组。收集两组患者性别、年龄、吸烟史、饮酒史、肿瘤直径、TNM分期、远处转移、淋巴结转移、局部侵犯、手术方式等一般资料。

1.3 统计学处理

应用SPSS 24.0统计学软件处理数据，计量资料以均数±标准差（\overline{x}\pm s）表示，采用t检验，计数资料以例数（百分比）[n（%）]表示，采用χ²检验，将具有统计学意义的因素进一步采用多因素Logistic回归分析，使用R3.5.3软件构建列阵图风险模型，采用Bootstrap法检验模型的准确性，采用受试者工作曲线（receiver operating characteristic curve，ROC）曲线评估预测效能。校验水准α=0.05。P<0.05为差异有统计学意义。

2 结　果

2.1 PDTC患者临床特征

55例PDTC患者中男性22例，女性33例，男女构成比例为1∶1.5；年龄28~76岁，平均（43.55±12.37）岁；肿瘤直径≥3 cm 25例，肿瘤直径<3 cm 30例；T1~T2分期28例，T3~T4分期27例；N0分期32例，N1分期23例；M0分期38例，M1分期17例（表1）。

2.2 PDTC患者预后情况

经过3年的随访，55例患者中16例患者术后肿瘤进展或死亡，具体包括，15例患者术后肿瘤进展，1例患者死亡（图1）。

2.3 PDTC患者预后不良的单因素分析

单因素分析显示，预后不良组肿瘤直径、TNM分期、局部侵犯、手术方式、脉管侵犯、神经受累与预后良好组比较，差异有统计学意义（均P<0.05）；两组在性别、年龄、吸烟史、饮酒史、远处转移、淋巴结转移、内分泌治疗、放疗、化疗等方面差异无统计学意义（均P>0.05）（表2）。

2.4 PDTC患者预后不良的多因素Logistic回归分析

以PDTC患者预后不良作为因变量（Y），将单因素分析筛选出有统计学意义的因素作为自变量（X）进行多因素Logistic回归分析，赋值结果见表3。结果显示，肿瘤直径、TNM分期、局部侵犯、手术方式、脉管侵犯、神经受累为PDTC患者预后不良的危险因素（P<0.05）（表4）。

2.5 PDTC患者预后不良的风险列线图模型构建

将肿瘤直径、TNM分期、局部侵犯、手术方式等作为预测变量进行PDTC患者预后不良的风险列线图模型构建（图2）。

2.6 PDTC患者预后不良风险列线图模型的验证

使用Bootstrap法内部验证结果显示，C-指数为0.794（95% CI=0.725~846），表明区分度良好（图3）。

2.7 PDTC患者预后不良风险列线图模型的预测价值评估

使用ROC曲线检验该风险列线图模型的预测效能，结果显示，ROC曲线下面积、敏感度、特异度分别为0.817、82.26%、86.35%（图4）。

3 讨　论

PDTC是一种罕见的甲状腺恶性肿瘤，其高度异质且预后差，是甲状腺癌研究领域的关键挑战之一^[9]。尽管PDTC发病率仅占所有甲状腺癌的0.5%~7%，但一些小样本研究显示，PDTC是导致滤泡细胞来源的甲状腺癌患者死亡的重要原因，具有较高的研究价值^[10-12]。受限于样本量和研究设计，PDTC的预后机制及风险分层策略仍不明确。本研究通过系统性分析PDTC患者的临床病理特征，并构建预后模型，以期为临床制定个体化治疗策略提供重要依据。

据报道^[13-14]，与分化型甲状腺癌相比，PDTC的侵袭性及局部复发倾向更强，且复发过程中常伴随着肿瘤的去分化，导致患者预后情况不甚理想。同时，一些研究^[15-17]也显示，部分PDTC患者对放射性碘治疗反应较差，因此，仅有1/3的患者在初次治疗后达到疾病缓解，多数PDTC患者在治疗后均需定期进行复查监测。可见，深入研究PDTC预后相关影响因素对于改善患者治疗效果及预后结局具有重要意义。肿瘤直径、TNM分期、局部侵犯等均是预测恶性肿瘤患者预后的常见指标^[18-20]。本研究通过对比不同预后结局PDTC患者的一般资料发现，肿瘤直径、TNM分期、局部侵犯是PDTC患者预后不良的危险因素，提示肿瘤直径、TNM分期、局部侵犯可能对该类患者预后结局造成影响。另外，本研究结果还显示，手术方式为PDTC患者预后不良的危险因素。说明手术方式的选择与PDTC患者预后结局有关。分析原因可能是，虽然次全切术可较大程度保留患者的甲状腺组织，对甲状旁腺的损伤较小，但全切术对患者癌变组织清除更为彻底，可降低发生病灶远处转移及局部复发的风险^[21-22]。甲状腺脉管侵犯是局部淋巴结转移和远处转移的独立预测因子，其发生机制可能与肿瘤细胞通过血管或淋巴管扩散至全身相关，导致术后残留和复发风险显著增加^[23-24]。神经受累（如侵犯喉返神经或颈交感神经）不仅会直接破坏周围正常组织功能，还可能通过神经浸润途径加速肿瘤的局部进展和远处转移^[25]。这两类病理特征的共存往往提示肿瘤具有高度侵袭性，且与甲状腺癌患者生存率下降密切相关^[26-27]。本研究结果也得到了类似的结论，脉管侵犯和神经受累是PDTC患者预后不良的危险因素。对PDTC而言，由于其恶性程度较高，因此脉管侵犯和神经受累的发生率也更高，更易促使肿瘤细胞通过血液、淋巴系统转移至远处器官，或沿神经束膜扩散导致邻近器官功能障碍，显著恶化了PDTC患者的生存情况和生活质量^[28-29]。

近年来，列线图模型在医学研究领域受到了越来越多的关注，其可将多个预测指标整合为可视化平面图形，帮助医师快速对患者病情进行判断和评价^[20]。本研究将多因素Logistic回归分析筛选出的危险因素作为预测变量构建PDTC患者预后列线图模型，使用Bootstrap法及ROC曲线检验显示，模型具有较好的预测效能和准确性。与传统的单纯使用TNM分期进行预后评估相比，本研究所建立的列线图模型纳入了更多影响因素，可帮助提升临床对PDTC患者预后评估的准确性。当然，本研究也存在一定的局限性，研究仅进行了为期3年的随访，时间较短，并且本研究所选取的样本量较小且具有区域性，可能导致结果产生偏差，需开展大样本的实验进行外部验证。

综上所述，肿瘤直径、TNM分期、局部侵犯、手术方式、脉管侵犯、神经受累等是影响PDTC患者预后不良的危险因素，基于以上因素构建的风险列线图模型具有较好的预测效能。

参考文献

[1] Cirello V, Gambale C, Nikitski AV, et al. Poorly differentiated thyroid carcinoma: molecular, clinico-pathological hallmarks and therapeutic perspectives[J]. Panminerva Med, 2024, 66(2):155-173. doi:10.23736/S0031-0808.23.05040-1 .

[2] Harahap AS, Roren RS, Imtiyaz S. A comprehensive review and insights into the new entity of differentiated high-grade thyroid carcinoma[J]. Curr Oncol, 2024, 31(6):3311-3328. doi:10.3390/curroncol31060252 .

[3] Violetis O, Konstantakou P, Spyroglou A, et al. The long journey towards personalized targeted therapy in poorly differentiated thyroid carcinoma (PDTC): a case report and systematic review[J]. J Pers Med, 2024, 14(6):654. doi:10.3390/jpm14060654 .

[4] Xu B, Ghossein RA. Advances in thyroid pathology: high grade follicular cell-derived thyroid carcinoma and anaplastic thyroid carcinoma[J]. Adv Anat Pathol, 2023, 30(1):3-10. doi:10.1097/PAP.0000000000000380 .

[5] Sun W, Wang Z, Xiang J, et al. Newly proposed survival staging system for poorly differentiated thyroid cancer: a SEER-based study[J]. J Endocrinol Invest, 2023, 46(5):947-955. doi:10.1007/s40618-022-01958-5 .

[6] Alam IS, Patel KN. Management of poorly differentiated thyroid cancer and differentiated high-grade thyroid carcinoma[J]. Surg Clin North Am, 2024, 104(4):751-765. doi:10.1016/j.suc.2024.02.005 .

[7] Prinzi A, van Velsen EFS, Belfiore A, et al. Brain metastases in differentiated thyroid cancer: clinical presentation, diagnosis, and management[J]. Thyroid, 2024, 34(10):1194-1204. doi:10.1089/thy.2024.0240 .

[8] Juhlin CC, Mete O, Baloch ZW. The 2022 WHO classification of thyroid tumors: novel concepts in nomenclature and grading[J]. Endocr Relat Cancer, 2022, 30(2):e220293. doi:10.1530/ERC-22-0293 .

[9] Wan Z, Wang B, Yao J, et al. Predictive factors and clinicopathological characteristics of outcome in poorly differentiated thyroid carcinoma: a single-institution study[J]. Front Oncol, 2023, 13:1102936. doi:10.3389/fonc.2023.1102936 .

[10] Kim JY, Myung JK, Kim S, et al. Prognosis of poorly differentiated thyroid carcinoma: a systematic review and meta-analysis[J]. Endocrinol Metab (Seoul), 2024, 39(4):590-602. doi:10.3803/EnM.2024.1927 .

[11] Chatterjee S, Mair M, Shaha AR, et al. Current evidences in poorly differentiated thyroid carcinoma: a systematic review and subsection meta-analysis for clinical decision making[J]. Endocrine, 2024, 85(2):509-519. doi:10.1007/s12020-024-03771-x .

[12] Cracolici V. No longer well-differentiated: diagnostic criteria and clinical importance of poorly differentiated/high-grade thyroid carcinoma[J]. Surg Pathol Clin, 2023, 16(1):45-56. doi:10.1016/j.path.2022.09.006 .

[13] Goodsell K, Ermer J, Amjad W, et al. External beam radiotherapy for thyroid cancer: Patients, complications, and survival[J]. Am J Surg, 2023, 225(6):994-999. doi:10.1016/j.amjsurg.2023.01.009 .

[14] Gu H, Wang J, Ran W, et al. Anaplastic and poorly differentiated thyroid carcinomas: genetic evidence of high-grade transformation from differentiated thyroid carcinoma[J]. J Pathol Clin Res, 2024, 10(2):e356. doi:10.1002/cjp2.356 .

[15] Cracolici V, Cipriani NA. High-grade non-anaplastic thyroid carcinomas of follicular cell origin: a review of poorly differentiated and high-grade differentiated carcinomas[J]. Endocr Pathol, 2023, 34(1):34-47. doi:10.1007/s12022-023-09752-6 .

[16] 张淼, 许丹晴, 周力, 等. 影响分化型甲状腺癌术后促甲状腺激素抑制治疗疗效因素的研究进展[J]. 中国普通外科杂志, 2024, 33(6):1012-1017. doi:10.7659/j.issn.1005-6947.2024.06.017 .

[17] Zhang M, Xu DQ, Zhou L, et al. Factors influencing the efficacy of thyrotropin suppression therapy after surgery for differentiated thyroid cancer:a review of research progress[J]. China Journal of General Surgery, 2024, 33(6):1012-1017. doi:10.7659/j.issn.1005-6947.2024.06.017 .

[18] Ryu H, Wu HG, Lee KE, et al. Effect of postoperative radiotherapy for patients with differentiated thyroid cancer[J]. Clin Endocrinol (Oxf), 2023, 98(6):803-812. doi:10.1111/cen.14865 .

[19] Thompson LDR. High grade differentiated follicular cell-derived thyroid carcinoma versus poorly differentiated thyroid carcinoma: a clinicopathologic analysis of 41 cases[J]. Endocr Pathol, 2023, 34(2):234-246. doi:10.1007/s12022-023-09770-4 .

[20] 闫军. 全甲状腺切除术对双侧甲状腺癌患者应激反应、喉返神经和甲状旁腺损伤的影响[J]. 陕西医学杂志, 2022, 51(7):831-834. doi:10.3969/j.issn.1000-7377.2022.07.015 .

[21] Yan J. The impact of total thyroidectomy on stress response, recurrent laryngeal nerve injury, and parathyroid gland damage in patients with bilateral thyroid carcinoma[J]. Shaanxi Medical Journal, 2022, 51(7):831-834. doi:10.3969/j.issn.1000-7377.2022.07.015 .

[22] 储冰峰, 胡瑛君. 甲状腺乳头状癌颈部淋巴转移的性别与年龄相关差异基因与通路分析[J]. 中国普通外科杂志, 2024, 33(11):1825-1834. doi:10.7659/j.issn.1005-6947.2024.11.009 .

[23] Chu BF, Hu YJ. Gender-and age-related differential genes and pathways in cervical lymph node metastasis of papillary thyroid carcinoma[J]. China Journal of General Surgery, 2024, 33(11):1825-1834. doi:10.7659/j.issn.1005-6947.2024.11.009 .

[24] Oba T, Maeno K, Kiyosawa N, et al. Neutrophil-to-lymphocyte ratio is associated with the proportion of poorly differentiated components in papillary thyroid carcinoma[J]. J Surg Res, 2023, 287:47-54. doi:10.1016/j.jss.2023.01.010 .

[25] Coca-Pelaz A, Rodrigo JP, Agaimy A, et al. Poorly differentiated thyroid carcinomas: conceptual controversy and clinical impact[J]. Virchows Arch, 2024, 484(5):733-742. doi:10.1007/s00428-024-03752-5 .

[26] Kim KM, Ahn AR, Hong YT, et al. Primary small cell thyroid carcinoma combined with poorly differentiated thyroid carcinoma, evidence for a common origin: a case report[J]. Oncol Lett, 2023, 25(6): 233. doi: 10.3892/ol.2023.13819 .

[27] Huang D, Zhang J, Zheng X, et al. Predictive value of hematologic parameters and clinicopathological features of poorly differentiated thyroid carcinoma and anaplastic thyroid carcinoma[J]. Langenbecks Arch Surg, 2024, 409(1):241. doi:10.1007/s00423-024-03431-8 .

[28] Alwelaie Y, Howaidi A, Tashkandi M, et al. Revisiting the cytomorphological features of poorly differentiated thyroid carcinoma: a comparative analysis with indeterminate thyroid fine-needle aspiration samples[J]. J Am Soc Cytopathol, 2023, 12(5):331-340. doi:10.1016/j.jasc.2023.05.002 .

[29] Liu Z, Xu Q, Xia H, et al. Dynamic estimates of survival of patients with poorly differentiated thyroid carcinoma: a population-based study[J]. Front Endocrinol (Lausanne), 2024, 15:1375274. doi:10.3389/fendo.2024.1375274 .

[30] Poma AM, Macerola E, Ghossein RA, et al. Prevalence of differentiated high-grade thyroid carcinoma among well-differentiated tumors: a systematic review and meta-analysis[J]. Thyroid, 2024, 34(3):314-323. doi:10.1089/thy.2023.0350 .

[31] 程若川, 刘文, 董治中. 局部晚期甲状腺癌的成因分析与临床启示[J]. 中国普通外科杂志, 2024, 33(11):1751-1758. doi:10.7659/j.issn.1005-6947.2024.11.001 .

[32] Cheng RC, Liu W, Dong ZZ. Etiological analysis and clinical implications of locally advanced thyroid cancer[J]. China Journal of General Surgery, 2024, 33(11):1751-1758. doi:10.7659/j.issn.1005-6947.2024.11.001 .

[33] Torous VF, Jitpasutham T, Baloch Z, et al. Cytologic features of differentiated high-grade thyroid carcinoma: a multi-institutional study of 40 cases[J]. Cancer Cytopathol, 2024, 132(8):525-536. doi:10.1002/cncy.22874 .