Introduction
Thyroid cancer is the most common endocrine malignancy, with a steadily increasing incidence worldwide.Reference Yin, Tang, Guo and Wen1 Most cases present as thyroid nodules, which are highly prevalent in the general population.Reference Dean and Gharib2 Although the majority of nodules are benign, 5–15 per cent carry a risk of malignancy.Reference Guth, Theune, Aberle, Galach and Bamberger3 Thyroid ultrasonography and histopathology are the main methods for malignancy assessment; however, limited accessibility and the invasive nature of procedures such as fine needle aspiration highlight the need for simple, cost-effective, non-invasive biomarkers.Reference Haugen, Alexander, Bible, Doherty, Mandel and Nikiforov4
Pre-operative inflammatory markers have been reported to differ between benign and malignant nodules and may relate to prognosis.Reference Kars, Şahin, Kılıç, Sakat and Bilen5, Reference Pagano, Mele, Sama, Zavattaro, Caputo and De Marchi6 Yet, the roles of novel indices, particularly haemoglobin–albumin–lymphocyte–platelet (‘HALP’) score and systemic inflammatory response index, in differentiating malignancy remain unclear.
Pre-operative inflammatory indices, particularly haemoglobin–albumin–lymphocyte–platelet score and systemic immune-inflammation index, have emerged as promising candidates for distinguishing malignant from benign thyroid lesions given their ability to integrate haematological and nutritional parameters from routine blood tests. These markers may complement conventional imaging and cytology, addressing the unmet need for non-invasive, cost-effective diagnostic tools.
Accordingly, this study aimed to evaluate the predictive performance of systemic inflammatory markers – including neutrophil-to-lymphocyte ratio, lymphocyte-to-monocyte ratio, platelet-to-lymphocyte ratio, systemic immune-inflammation index, systemic inflammatory response index and the haemoglobin–albumin–lymphocyte–platelet score – in differentiating malignant from benign thyroid lesions.
Materials and methods
Study design and population
This retrospective study included patients aged 18 years or more who underwent partial or total thyroidectomy between 2015 and 2025 at the ENT and general surgery departments of a tertiary care university hospital. Clinical, laboratory and histopathological data, including pre-operative complete blood counts, were obtained from electronic records, operative logs and medical charts.
Patients were classified as having benign or malignant thyroid lesions based on histopathology, according to the World Health Organization Classification of Endocrine and Neuroendocrine Tumours (5th edition).7 To avoid potential misclassification, low-risk thyroid neoplasms that could not be clearly categorised as benign or malignant – namely, non-invasive follicular thyroid neoplasm with papillary-like nuclear features, follicular tumour of uncertain malignant potential and well-differentiated tumour of uncertain malignant potential – were excluded from the study.
Data collection and calculation of inflammatory indices
Inflammatory markers, including the neutrophil-to-lymphocyte ratio, lymphocyte-to-monocyte ratio, platelet-to-lymphocyte ratio, systemic immune-inflammation index, systemic inflammatory response index and haemoglobin–albumin–lymphocyte–platelet score, were calculated using data obtained from blood samples collected during routine pre-operative preparation. Samples were obtained from patients in a fasting state. Patients with concurrent inflammatory or infectious conditions were excluded, as such conditions could influence systemic inflammatory indices and preclude surgery. Systemic immune-inflammation index and systemic inflammatory response index were calculated using the following formulas: platelet count × neutrophil count / lymphocyte count, and neutrophil count × monocyte count / lymphocyte count, respectively. The haemoglobin–albumin–lymphocyte–platelet score was calculated using the formula: haemoglobin (g/l) × albumin (g/l) × lymphocyte count (/l) / platelet count (/l).
Ethics
The study protocol was approved by the Kırıkkale University Non-Interventional Research Ethics Committee (date: 10 December 2025; decision number: 2025.12.15). Written informed consent was obtained from all patients.
Statistical analysis
Statistical evaluation of the data was performed using the computer-based statistical software package IBM SPSS Statistics, version 21.0. The normality of distribution for continuous variables was assessed using both visual methods (histograms and probability plots) and analytical methods (Kolmogorov–Smirnov or Shapiro–Wilk tests). Descriptive statistics were expressed as mean ± standard deviation for continuous variables that met the assumption of normal distribution, and as median with interquartile range (25th–75th percentile) for non-normally distributed continuous variables. Categorical variables were presented as frequency (n) and percentage. Comparisons of continuous variables between benign and malignant groups were performed using the independent-samples t-test for normally distributed data and the Mann–Whitney U test for non-normally distributed data (inflammatory indices). The chi-square test was used to analyse categorical variables (e.g. sex) between groups.
Receiver operating characteristic curve analysis was performed to evaluate the diagnostic performance of inflammatory indices, including neutrophil-to-lymphocyte ratio, lymphocyte-to-monocyte ratio, platelet-to-lymphocyte ratio, systemic immune-inflammation index, systemic inflammatory response index and haemoglobin–albumin–lymphocyte–platelet score, in predicting thyroid malignancy, as well as to determine area under the curve values and optimal cut-off points. Sensitivity and specificity values at the identified cut-off levels were calculated. Correlations among inflammatory indices were evaluated using Spearman’s rank correlation analysis. To minimise the risk of multicollinearity that could arise from strong correlations among indices and to ensure statistical stability of the models, the independent association of each index with malignancy was examined using separate multivariable logistic regression models. Potential confounding factors, including age and sex, were included in all baseline models, allowing the independent predictive value of each inflammatory index for malignancy to be assessed after adjustment for other related parameters.
In all statistical analyses, a two-tailed p-value of less than 0.05 was considered statistically significant.
Results and analysis
A total of 316 patients who underwent thyroidectomy were included in the study. Of these, 186 patients were diagnosed with benign thyroid pathologies and 130 with malignant thyroid pathologies. Demographic characteristics of the study population are summarised in Table 1. Patients with benign and malignant thyroid pathologies differed significantly with respect to age and sex distribution (Table 1).
Demographic characteristics and systemic inflammatory indices for benign versus malignant thyroid pathologies
Table 1 Long description
Demographics and systemic inflammatory indices are compared between benign thyroid pathology (186 people) and malignant pathology (130 people), with p-values indicating which group differences are statistically significant. Malignant cases are younger on average (about 52.6 years) than benign cases (about 56.4 years), and the sex distribution differs (malignant 104 female and 26 male; benign 130 female and 56 male). The haemoglobin to albumin to lymphocyte to platelet score is lower in malignant cases (about 49.6) than benign cases (about 64.4), showing a significant difference. Several inflammation-related ratios are higher in malignant disease: neutrophil-to-lymphocyte ratio median about 2.07 versus 1.76, platelet-to-lymphocyte ratio median about 126.1 versus 102.6, and systemic immune-inflammation index median about 587.1 versus 476.9; all are statistically significant. The lymphocyte-to-monocyte ratio is slightly lower in malignant cases (median about 5.42) than benign cases (median about 5.69) and is statistically significant. The systemic inflammatory response index is similar between groups (medians about 0.87 malignant and 0.82 benign) and is not statistically significant. Values are reported as means with standard deviations for some measures and as medians with interquartile ranges for others, so comparisons should consider the different summary types.
* Data represent median (25th–75th percentile). †Indicates statistical significance (p < 0.05). SD = standard deviation
Compared with the benign group, patients with malignant thyroid lesions had significantly lower haemoglobin–albumin–lymphocyte–platelet score and lymphocyte-to-monocyte ratio values and significantly higher neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio and systemic immune-inflammation index values, whereas systemic inflammatory response index did not differ significantly between the groups (Table 1).
The distribution of benign and malignant thyroid pathologies is summarised in Table 2. Papillary thyroid carcinoma constituted the majority of malignant lesions, whereas Hashimoto’s thyroiditis (chronic lymphocytic thyroiditis) was the most common benign pathology (Table 2).
Distribution of benign and malignant thyroid pathologies according to frequency
Table 2 Long description
Counts and percentages of thyroid pathology subtypes in a study population of 316 cases are reported, split into benign and malignant groups. Benign lesions account for 186 cases, representing 58.9 percent of all cases, while malignant lesions account for 130 cases, representing 41.1 percent. Within benign lesions, Hashimotos thyroiditis is most common with 62 cases, followed by nodular hyperplasia with 57 and multinodular goitre with 42; these correspond to 19.6 percent, 18.0 percent, and 13.3 percent of the total population, respectively. Less frequent benign subtypes are follicular adenoma with 16 cases and Hurthle cell adenoma with 9 cases. Within malignant lesions, papillary thyroid carcinoma dominates with 110 cases, making up 84.6 percent of malignancies and 34.8 percent of the total population. Other malignant subtypes are much less common: follicular carcinoma has 12 cases, medullary carcinoma and Hurthle cell carcinoma have 3 cases each, and poorly differentiated carcinoma has 2 cases. Percentages are presented both within each group and across the full study population, so within-group values should not be compared directly to overall values.
Note: Within-group percentages indicate the proportion of each subtype within the benign or malignant categories, while overall percentages represent their distribution in the total study population (n = 316). Borderline thyroid neoplasms – non-invasive follicular thyroid neoplasm with papillary-like nuclear features, follicular tumour of uncertain malignant potential, and well-differentiated tumour of uncertain malignant potential – were excluded from the analysis and are not presented in the table.
Correlation analysis demonstrated moderate-to-strong intercorrelations among the inflammatory indices, supporting the use of separate multivariable regression models to minimise multicollinearity (Table 3).
Correlation analysis between systemic inflammatory indices
Table 3 Long description
Spearman rank correlations compare six systemic inflammatory indices: HALP, NLR, SII, SIRI, PLR, and LMR. The strongest positive relationship is between HALP and PLR, with a very high correlation and p below 0.001. NLR is strongly correlated with SII and also strongly correlated with SIRI, both with p below 0.001. SIRI shows a strong positive correlation with LMR, with p below 0.001. SII has moderate to strong positive correlations with HALP and PLR, each with p below 0.001. The weakest relationship is between HALP and SIRI, which is small and not statistically significant with p 0.067. All other reported pairwise correlations are statistically significant, including the small but significant correlation between PLR and SIRI with p 0.005. Correlations indicate association only and do not establish causation.
Note: Correlation co-efficients were calculated using Spearman’s rank correlation analysis. The r values represent Spearman’s rho.
* Indicates statistical significance (p < 0.05). HALP = haemoglobin–albumin–lymphocyte–platelet score; NLR = neutrophil-to-lymphocyte ratio; SII = systemic immune-inflammation index; SIRI = systemic inflammatory response index; PLR = platelet-to-lymphocyte ratio; LMR = lymphocyte-to-monocyte ratio
In order to minimise the risk of multicollinearity, each inflammatory index was evaluated in a separate multivariable logistic regression model adjusted for age and sex (Table 4). Haemoglobin–albumin–lymphocyte–platelet score and systemic immune-inflammation index remained independent predictors of malignancy, whereas neutrophil-to-lymphocyte ratio was not significant and lymphocyte-to-monocyte ratio showed borderline significance.
Comparison of multivariable logistic regression models for thyroid malignancy
Table 4 Long description
Four separate multivariable logistic regression models report adjusted odds ratios with confidence intervals and p-values for thyroid malignancy, each including age and sex plus one inflammatory index. Age shows a small inverse association across models, with p-values near but not below the significance threshold. Sex is not significant in Models 1 to 3, but is significant in Model 4 with higher odds for the listed sex category. In Model 1, higher HALP score is significantly associated with lower odds of malignancy, with a p-value below 0.001. In Model 3, higher SII is significantly associated with higher odds, with a p-value of 0.035. NLR in Model 2 and LMR in Model 4 are not statistically significant, though LMR is borderline. Each inflammatory index is modeled separately, so results should be compared cautiously across models rather than interpreted as simultaneous independent effects.
Note: Model 1: age, sex and haemoglobin–albumin–lymphocyte–platelet (HALP) score. Model 2: age, sex and neutrophil-to-lymphocyte ratio (NLR). Model 3: age, sex and systemic immune-inflammation index (SII). Model 4: age, sex and lymphocyte-to-monocyte ratio (LMR). Due to strong correlations among inflammatory indices, each index was evaluated in separate multivariable models to avoid multicollinearity. Adjusted odds ratio (OR) values were controlled for age and sex. *Indicates statistical significance (p < 0.05). CI = confidence interval
Receiver operating characteristic curve analysis demonstrated that platelet-to-lymphocyte ratio and haemoglobin–albumin–lymphocyte–platelet score had the highest discriminative performance among the evaluated inflammatory indices, although the overall diagnostic accuracy remained modest (Figure 1 and Table 5). Detailed area under the curve values, optimal cut-off values, sensitivity and specificity are presented in Table 5.
Receiver operating characteristic curves of systemic inflammatory indices for predicting thyroid malignancy. HALP = haemoglobin–albumin–lymphocyte–platelet score; AUC = area under the curve; NLR = neutrophil-to-lymphocyte ratio; LMR = lymphocyte-to-monocyte ratio; PLR = platelet-to-lymphocyte ratio; SII = systemic immune-inflammation index; SIRI = systemic inflammatory response index.
Figure 1 Long description
A multi-line graph showing receiver operating characteristic (ROC) curves for six systemic inflammatory indices used to distinguish malignant from benign thyroid pathologies. The horizontal axis represents 1–specificity, and the vertical axis represents sensitivity, both ranging from 0.0 to 1.0, with tick marks at intervals of 0.2. Six color-coded, stepwise ROC curves are displayed: HALP (blue, AUC = 0.689), NLR (green, AUC = 0.400), LMR (light grey, AUC = 0.567), PLR (purple, AUC = 0.696), SII (yellow, AUC = 0.630), and SIRI (red, AUC = 0.511). A diagonal reference line extends from the lower-left corner (0.0, 0.0) to the upper-right corner (1.0, 1.0), representing no discriminative ability (AUC = 0.5). The curves show different patterns of sensitivity and specificity across the plotted range.
Receiver operating characteristic analysis of inflammatory indices for predicting malignancy
Table 5 Long description
Receiver operating characteristic results compare six inflammatory indices for predicting malignancy, reporting area under the curve with confidence intervals, cut-offs, p-values, sensitivity, and specificity. PLR performs best with AUC 0.696, sensitivity 67.7 percent, and specificity 62.4 percent at a cut-off above 109.53, with a statistically significant p-value. HALP is similar with AUC 0.689, sensitivity 64.5 percent, and specificity 68.5 percent at a cut-off below 55.87, also statistically significant. SII shows moderate discrimination with AUC 0.630 and sensitivity and specificity near 58 percent at a cut-off above 522.49, statistically significant. LMR is weaker with AUC 0.567 and sensitivity and specificity in the mid 40 percent range at a cut-off below 5.543, but still statistically significant. SIRI is near chance with AUC 0.511 and a non-significant p-value, while NLR performs poorly with AUC 0.400 and low sensitivity and specificity despite statistical significance. Cut-offs were selected to balance sensitivity and specificity, so performance may differ in other samples or clinical settings.
Note: Cut-off values were determined using receiver operating characteristic curve analysis based on the Youden index. *Indicates statistical significance (p < 0.05). AUC = area under the curve; CI = confidence interval; HALP = haemoglobin–albumin–lymphocyte–platelet score; NLR = neutrophil-to-lymphocyte ratio; LMR = lymphocyte-to-monocyte ratio; PLR = platelet-to-lymphocyte ratio; SII = systemic immune-inflammation index; SIRI = systemic inflammatory response index
Discussion
Papillary thyroid carcinoma is associated with a favourable prognosis, with reported 10-, 15- and 20-year survival rates of 97 per cent, 95 per cent and 90 per cent, respectively.Reference Ito, Miyauchi, Kihara, Fukushima, Higashiyama and Miya8 Therefore, survival analysis based on parameters such as the haemoglobin–albumin–lymphocyte–platelet score was beyond the scope of the present study. Instead, the primary aim of this study was to evaluate the diagnostic performance of systemic inflammatory indices in differentiating malignant from benign thyroid lesions, particularly during the pre-operative evaluation of patients with indeterminate findings.
The potential diagnostic relevance of these indices is supported by the established link between systemic inflammation and tumour development. Inflammatory cell infiltration within the tumour microenvironment may exert bidirectional effects on tumour progression: lymphocytes can suppress tumour invasion and metastasis through anti-tumoural immune responses, whereas neutrophils and platelets may promote invasion, angiogenesis and metastatic dissemination.Reference Masson-Lecomte, Rava, Real, Hartmann, Allory and Malats9, Reference Mantovani, Allavena, Sica and Balkwill10 Peripheral blood parameters reflecting this systemic inflammatory response have been associated with cancer development and disease course.Reference Eo, Jeong, Chang, Won, Choi and Kim11–Reference Olsson and Cedervall13 Accordingly, indices such as neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio and lymphocyte-to-monocyte ratio are considered practical and accessible biomarkers reflecting the balance between pro-tumour inflammation and anti-tumoural immunity.Reference Chanmee, Ontong, Konno and Itano14–Reference Fu, Liu, Zhang, Liu and Wang16 Elevated neutrophil-to-lymphocyte ratio and platelet-to-lymphocyte ratio levels have been linked to adverse clinical features in several solid tumours, whereas lymphocyte-to-monocyte ratio has shown an inverse relationship with these markers.Reference Grimm, Rieth, Hoefert, Krimmel, Rieth and Teriete17
Large-scale prospective cohort studies have demonstrated significant associations between systemic inflammatory markers and cancer risk, reporting positive correlations for systemic immune-inflammation index, neutrophil-to-lymphocyte ratio and platelet-to-lymphocyte ratio, and a negative correlation for lymphocyte-to-monocyte ratio.Reference Nøst, Alcala, Urbarova, Byrne, Guida and Sandanger18 Consistent with these findings, our study also demonstrated similar associations between systemic inflammatory indices and malignancy. Among these markers, systemic immune-inflammation index appears to have a particularly strong association with cancer risk in solid tumours.Reference Nøst, Alcala, Urbarova, Byrne, Guida and Sandanger18 Pre-operative systemic immune-inflammation index levels have also been reported to aid in differentiating malignant from benign lesions in patients with papillary thyroid carcinoma.Reference Vural, Muhtaroğlu and Güngör19 In addition, predictive models incorporating systemic immune-inflammation index and lymphocyte-to-monocyte ratio have been proposed to improve diagnostic accuracy during the pre-operative evaluation of thyroid nodules.Reference Tang, Li and Luo20
Consistent with previous reports, systemic inflammatory indices have been suggested as potential biomarkers in thyroid cancer. Refaat et al. reported significantly higher platelet-to-lymphocyte ratio, systemic immune-inflammation index and systemic inflammatory response index levels in patients with malignant thyroid disease compared with controls,Reference Refaat, Eissa, Elnaggar, Mehesen, Eissa and Kamal21 and similar findings have been described in other studies evaluating the diagnostic value of inflammatory indices in thyroid malignancies.Reference Kars, Şahin, Kılıç, Sakat and Bilen5, Reference Ozmen, Timur, Calik, Altinkaynak, Simsek and Gozcu22 In line with these data, our study demonstrated significantly lower lymphocyte-to-monocyte ratio levels in the malignant group, supporting the inverse association reported in the literature.Reference Refaat, Eissa, Elnaggar, Mehesen, Eissa and Kamal21, Reference Offi, Romano, Cangiano, Filograna Pignatelli, Candela and Docimo23 Although neutrophil-to-lymphocyte ratio differed significantly between the malignant and benign groups, it was not identified as an independent predictor of malignancy in the multivariable analysis, suggesting that neutrophil-to-lymphocyte ratio alone may have limited discriminative value in this setting.Reference Ozmen, Timur, Calik, Altinkaynak, Simsek and Gozcu22, Reference Ceylan, Kumanlıoğlu, Oral, Ertan and Özcan24
Among systemic inflammatory indices, the haemoglobin–albumin–lymphocyte–platelet score merits particular attention due to its composite structure, incorporating haemoglobin, albumin, lymphocyte and platelet levels to reflect both inflammatory and nutritional status. Recently, Azad and Erman evaluated the diagnostic performance of the haemoglobin–albumin–lymphocyte–platelet score in patients with thyroid nodules, and reported significantly lower haemoglobin–albumin–lymphocyte–platelet score values in malignant lesions compared with benign nodules, with a moderate discriminative performance (area under the curve = 0.672).Reference Azad and Erman25 Our findings are consistent with this association and further strengthen the available evidence by externally validating the relationship between haemoglobin–albumin–lymphocyte–platelet score and thyroid malignancy in a larger surgical cohort while simultaneously comparing haemoglobin–albumin–lymphocyte–platelet score with multiple systemic inflammatory indices. Although previous studies have generally supported the association between lower haemoglobin–albumin–lymphocyte–platelet score and thyroid malignancy, differences in the reported diagnostic performance may reflect variations in study design, sample size, patient selection and clinicopathological characteristics of the study populations.
Haemoglobin–albumin–lymphocyte–platelet score has been shown to have prognostic value in various types of cancer.Reference Farag, Antar, Akosman, Ng and Whalen26 In various malignancies, low haemoglobin–albumin–lymphocyte–platelet scores have been linked to adverse clinicopathological features, including larger tumour size, advanced tumour–node–metastasis stage and lymphovascular invasion,Reference Zhai, Chen, Wu, Yang, Guo and Shao27, Reference Peng, Zhang, Tang, Zhang, Yang and Yu28 as well as poorer prognosis in gastric and hepatocellular cancers.Reference Chen, Xue, Wang, Chen, Zhang and Liu29, Reference Toshida, Itoh, Kayashima, Nagao, Yoshiya and Tomino30 Notably, patients with low haemoglobin–albumin–lymphocyte–platelet score have been reported to have an approximately fivefold increased risk of mortality, with lymphopenia – a component of haemoglobin–albumin–lymphocyte–platelet score – identified as an independent adverse prognostic factor across cancer types.Reference Wu, Oduyebo, Cobb, Cholakian, Kong and Fader31
Previous studies have highlighted the prognostic and diagnostic potential of the haemoglobin–albumin–lymphocyte–platelet score in various head and neck malignancies. Wu et al. reported that, among 319 predominantly male patients with pharyngeal cancers, low haemoglobin–albumin–lymphocyte–platelet score was associated with worse overall survival and remained an independent predictor in multivariate analysis.Reference Wu, Lin, Lo, Cheng, Hsu and Chen32 Similarly, Ogasawara et al. found that low haemoglobin–albumin–lymphocyte–platelet score significantly increased the risk of metachronous tumour development after endoscopic submucosal dissection in 144 patients with superficial pharyngeal squamous cell carcinoma.Reference Ogasawara, Kikuchi, Tanaka, Ochiai, Okamura and Hayasaka33 In laryngeal lesions, haemoglobin–albumin–lymphocyte–platelet score differed significantly only between squamous cell carcinoma and benign groups, showing limited utility in predicting high-grade dysplasia.Reference Çanakcı, Tulacı, Özden, Arslan, Yazıcı and Tulacı34 In our cohort, haemoglobin–albumin–lymphocyte–platelet score was significantly lower in the malignant group and emerged as an independent predictor of malignancy, suggesting its potential role as a complementary biomarker in differentiating malignant from benign thyroid lesions.
The optimal haemoglobin–albumin–lymphocyte–platelet score cut-off (less than 55.87) in this study was determined using receiver operating characteristic curve analysis and the Youden index, which maximises combined sensitivity and specificity for distinguishing malignant from benign thyroid lesions. While this threshold provided the best statistical discrimination within our cohort, it should not be considered an absolute biological boundary, as haemoglobin–albumin–lymphocyte–platelet score is influenced by haemoglobin, albumin, lymphocyte and platelet levels, nutritional status, systemic inflammation, and population-specific factors. The identified cut-off aligns with thresholds reported for other malignancies and vascular conditions (range, 40–60), reflecting reduced immune-nutritional reserve and increased chronic inflammatory burden.Reference Farag, Antar, Akosman, Ng and Whalen26, Reference Chen, Xue, Wang, Chen, Zhang and Liu29, Reference Toshida, Itoh, Kayashima, Nagao, Yoshiya and Tomino30 Overall, these findings suggest that inflammatory indices, including haemoglobin–albumin–lymphocyte–platelet score, may serve as adjunctive tools in the pre-operative evaluation of thyroid nodules, supporting risk stratification and potentially reducing unnecessary surgical interventions, particularly in patients with indeterminate or borderline cytology. However, the observed discriminative performance was only modest, and these biomarkers should not be used as stand-alone diagnostic tools. Instead, they should be interpreted in conjunction with established diagnostic modalities, including ultrasonography, Thyroid Imaging Reporting and Data System (‘TI-RADS’) classification, and fine needle aspiration cytology.
From a clinical perspective, systemic inflammatory indices, including haemoglobin–albumin–lymphocyte–platelet score and systemic immune-inflammation index, may serve as cost-effective adjuncts in the pre-operative evaluation of thyroid nodules, particularly in patients with indeterminate cytology (Bethesda classifications III or IV) or equivocal imaging findings. It should be emphasised that these markers are intended as adjunctive tools and do not replace standard diagnostic modalities such as cytology or Thyroid Imaging Reporting and Data System based imaging assessment. Given their modest diagnostic accuracy, their greatest clinical value lies in complementing existing diagnostic algorithms rather than guiding clinical decision-making independently. While these markers cannot replace cytology or histopathology, their integration into existing diagnostic algorithms may enhance risk stratification and support surgical decision-making. Interpretation should be cautious, given the variability of inflammatory markers across different clinical conditions, and always considered alongside comprehensive clinical assessment. This study benefits from a relatively large sample size and simultaneous evaluation of multiple inflammatory indices in histopathologically confirmed thyroid lesions. Multivariable analyses accounted for potential confounding effects of age and sex and helped reduce multicollinearity among parameters, as demographic differences between benign and malignant groups could influence haematological indices such as haemoglobin, platelet count and lymphocyte levels.
The retrospective single-centre design and reliance on single pre-operative laboratory measurements may limit the generalisability of our findings and may not fully capture dynamic inflammatory status. Transient conditions such as acute infections or short-term haematological changes could act as potential confounders, and longitudinal monitoring may better reflect chronic inflammation in future studies. Additionally, Hashimoto’s thyroiditis, present in 33.3 per cent of benign lesions, may influence systemic inflammatory markers, including haemoglobin–albumin–lymphocyte–platelet score, neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio and systemic immune-inflammation index, potentially contributing to variation within the benign group. Nonetheless, these cases were included due to the sample size and the study’s primary aim of evaluating overall discriminative performance, while future studies may consider subgroup analyses to further assess the impact of autoimmune thyroiditis. Other factors such as smoking status, diabetes mellitus, autoimmune diseases and chronic inflammatory conditions may also affect these inflammatory indices. Because these variables were not consistently available in this retrospective dataset, they could not be included in the multivariable regression models and should be considered potential residual confounders. Furthermore, because only patients who underwent thyroidectomy were included, our cohort represents a surgically selected population with a relatively high pre-test probability of malignancy. This may introduce selection bias and limit the generalisability of our findings to the broader population of patients with thyroid nodules managed in out-patient practice.
• Five systemic inflammatory indices differed significantly between benign and malignant thyroid pathologies
• Haemoglobin–albumin–lymphocyte–platelet score was lower in malignant lesions
• Neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio and systemic immune-inflammation index were higher in malignant lesions
• Haemoglobin–albumin–lymphocyte–platelet score and systemic immune-inflammation index were independent predictors of malignancy
• Platelet-to-lymphocyte ratio and haemoglobin–albumin–lymphocyte–platelet score showed the highest, albeit modest, diagnostic performance for differentiating malignancy
Conclusion
Several systemic inflammatory indices, particularly haemoglobin–albumin–lymphocyte–platelet score and systemic immune-inflammation index, were independently associated with thyroid malignancy after adjustment for age and sex. Platelet-to-lymphocyte ratio and haemoglobin–albumin–lymphocyte–platelet score showed modestly higher discriminative ability in receiver operating characteristic analysis compared with other indices. These readily available biomarkers may provide supportive information in the pre-operative differentiation of benign and malignant thyroid lesions and aid in risk stratification. Further prospective studies are needed to confirm their clinical utility.
Data availability statement
The data used to support the findings of this study are available from the corresponding author upon reasonable request.
Competing interests
None declared.
Ethical standards
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional guidelines on human experimentation (Kırıkkale University Non-Interventional Research Ethics Committee (date: 10 December 2025; decision number: 2025.12.15)) and with the Helsinki Declaration of 1975, as revised in 2008.
Open access
