The value of dual-energy computed tomography quantitative parameters in differentiating the sensitivity of small lymph nodes to chemoradiotherapy in nasopharyngeal carcinoma

Abstract

Purpose

To evaluate the value of dual-energy computed tomography (DECT) quantitative parameters in differentiating the sensitivity of small lymph nodes (short-axis diameter 5–9 mm) that are negative on magnetic resonance imaging (MRI) to chemoradiotherapy in patients with nasopharyngeal carcinoma (NPC).

Methods

Newly diagnosed NPC patients underwent DECT before treatment. The reference standard was the radiographic regression status of lymph nodes at 1 year after radiotherapy. They were classified into the effective disease (ED) group and the stable disease (SD) group based on their sensitivity to chemoradiotherapy. The characteristics of small lymph nodes, DECT parameters, and diagnostic value were compared.

Results

One year after radiotherapy, 65 and 60 small lymph nodes were detected in the ED and SD groups, respectively; the proportion of small lymph nodes in the lower neck was higher in the ED group than in the SD group, and all such nodes achieved complete response (CR). Normalized iodine concentration (NIC), and slope of the energy spectrum (λHU) were independent risk factors for differentiating the sensitivity of small lymph nodes to chemoradiotherapy (P < 0.05). In the venous phase, λHU demonstrated the strongest diagnostic performance among all DECT parameters, with an area under the curve (AUC) of 0.803 and an optimal cutoff value of 4.465. The diagnostic power of the combined parameters was higher, especially when the short-axis diameter of small lymph nodes was 7–9 mm, the location was in the lower neck, and λHU was > 4.46 in the venous phase; the diagnostic power AUC of the combined parameters reached 0.870.

Conclusion

DECT quantitative parameters can effectively assess the sensitivity to chemoradiotherapy of MRI-negative small lymph nodes in NPC and aid in evaluating metastatic risk. λHU in the venous phase is the most reliable single DECT parameter. Combining short-axis diameter and nodal region provides an optimized reference for metastatic risk assessment of small lymph nodes.

Introduction

The N stage is a high-risk prognostic factor for distant metastasis and local recurrence in patients with nasopharyngeal carcinoma (NPC) [1, 2]. Local recurrence is common among patients undergoing radical radiation therapy. Accurate evaluation of metastatic and non-metastatic cervical lymph nodes is essential for assessing treatment strategies, determining target areas and radiation doses, and predicting survival outcomes [3, 4].

Lymph node metastasis of NPC is primarily a regional metastasis; however there is a risk of jump metastasis, with an incidence rate of 0.5%–7.9%, and the distribution is not absolutely orderly [5, 6]. More than 70% of patients with NPC have cervical lymph node involvement at diagnosis, with retropharyngeal lymph nodes first, followed by lymph nodes in the upper neck (II, III, or Va regions), and finally metastasis to the lower neck (IV and Vb regions) [7]. The risk of lymph node metastasis in areas I and VI is very low [8]. Magnetic resonance imaging (MRI) is the current standard imaging method for N staging of NPC, and diagnostic criteria for cervical lymph node metastasis on MRI include [9, 10]: (1) minimal diameter of ≥ 10 mm on maximum transverse images; (2) central necrosis or enhanced margin of lymph nodes; (3) aggregation of three or more lymph nodes in high-risk areas, with a short-axis diameter of at least 8–10 mm for the single node; and (4) lymph nodes of any size with signs of capsular invasion, including unclear lymph node margins, irregular lymph node capsule enhancement, and infiltration into adjacent fat or muscle. However, in clinical practice, a considerable proportion of small lymphl nodes with occult micro-metastases do not meet these morphological criteria. It is hard to determine whether small lymph nodes with a short-axis diameter < 10 mm and without central necrosis, rim enhancement, or capsular invasion are truly metastatic. In addition to the MRI diagnostic criteria for post-pharyngeal and cervical lymph node metastases outlined in the international consensus guidelines [11], lymph nodes with high sensitivity to chemoradiotherapy and that show significant regression after chemoradiotherapy should also be considered as positive lymph nodes. Studies have shown that patients without cervical lymph node metastases are 7% more likely to develop distant metastases, and patients with more than three lymph nodes are 50% more likely to develop distant metastases [12]. Radiotherapy guidelines for metastatic lymph nodes recommend a dose of 66–70 Gy, whereas prophylactic irradiation dose is typically delivered at 50–54 Gy. Therefore, accurate evaluation of cervical lymph node status is important for target mapping and treatment decision-making in NPC radiotherapy.

Multimodal imaging such as MRI, positron emission tomography/computed tomography (PET/CT), and ultrasound can provide information on proton signal, metabolic activity, tissue density, and elasticity, improving the ability to identify metastatic lymph nodes. However, the performance of morphology-based or single functional parameters remains insufficient for evaluating micro-metastases in small lymph nodes, highlighting the need for novel techniques to overcome the limitations of individual modalities [13, 14]. Dual-energy computed tomography (DECT) provides detailed and accurate diagnostic information by simultaneously acquiring X-ray images at different energy levels. DECT can produce several sets of post-processing images, such as iodine maps, virtual single energy images, and effective atomic number maps [15], which can help identify the tissue composition, hemodynamics, and metabolic activity of tumors. Several studies have demonstrated the usefulness of DECT-derived iodinographic quantitative parameters in differentiating metastatic and non-metastatic lymph nodes [16–20]; however, no studies have explored their value in differentiating the sensitivity of MRI-negative small lymph nodes (5–9 mm) to chemoradiotherapy in NPC.

Given the need for accurate staging and individualized therapy in current tumor treatment, this study used lymph node regression after chemoradiotherapy as the reference standard to explore the diagnostic utility of DECT quantitative parameters in distinguishing the sensitivity of MRI-negative small lymph nodes (5–9 mm) to chemoradiotherapy in patients with newly diagnosed NPC.

Materials and methods

Patient population

Analysis was undertaken of patients newly diagnosed with NPC from September 2021 to July 2024 who received pretreatment DECT and who met the following inclusion criteria: (1) pathologically confirmed NPC; (2) no radiotherapy, surgery or other anti-tumor treatment before DECT examination; (3) no history of iodine allergy or hyperthyroidism symptoms; (4) lymph nodes with a short-axis diameter of > 5 and < 10 mm visible on MRI or CT images of the neck; and (5) lymph nodes classified as MRI-negative. The exclusion criteria were as follows: (1) retropharyngeal lymph nodes; (2) incomplete clinical data; (3) poor image quality such that qualitative and quantitative analyses could not be performed; and (4) a history of tuberculosis or lymphoma.

All patients underwent DECT and enhanced MRI prior to treatment. The diagnostic criteria for small lymph nodes classified as negative on MRI were: (1) scattered lymph nodes with short-axis diameter ≥ 5 mm and < 10 mm; (2) no clustering, central necrosis or marginal enhancement; and (3) no signs of extracapsular invasion on T2-weighted images.

DECT examination and clinical treatment

All the patients were scanned using a DECT scanner (SOMATOM Definition Flash CT; Siemens Healthineers, Erlangen, Germany) and received radical intensity-modulated radiation therapy (IMRT) combined with induction and concurrent chemotherapy. The dose of radical radiotherapy for metastatic lymph nodes was 66–70 Gy, and the dose of prophylactic radiation for the neck was 50–54 Gy. The procedures for DECT examination and treatment were essentially consistent with those reported in our previous study [21].

DECT image analysis

The original image data was transferred with a reconstruction slice thickness of 1 mm and a slice spacing of 1 mm to a picture archiving and communication system and syngo.via a post-processing workstation (Siemens Healthineers). Open low-energy (90 keV sequence) and high-energy (150 keV sequence) images were collected simultaneously in CT dual-energy mode, and the images were viewed in Liver VNC, Single Energy +, and Rho/Z modes. The region of interest (ROI) delineation strictly follows the standardized method: under the single energy conditions of 40–140 keV and the best signal-to-noise ratio, the largest cross-section of each node was selected, and ROIs were completely outlined within the solid parenchyma, avoiding marginal heterogeneous tissues, measure the CT value, and select three different layers of the lesion to measure once, the average value is taken as the final result. The ROI was plotted on the common carotid artery of the same layer, and the focal iodine concentration (IC) value and common carotid artery IC values were measured. The effective atomic number (Zeff), linear mixed image with mixing ratio of 0.6 (Mix-0.6), and normalized IC (NIC; IC [lesion]/IC [common carotid artery]) were recorded; slope of the energy spectrum (λHU; [40–70 keV CT value]/30) (Fig. 1).

Fig. 1.

Dual-energy computed tomography images of a patient with nasopharyngeal carcinoma. Note: The patient with undifferentiated nonkeratinizing nasopharyngeal carcinoma (T4N1M0, IVa stage, AJCC/UICC 8th), small lymph node 1.0×0.7cm in the right neck level II. A Is an iodine-based substance diagram in the arterial phase, IC=2.5mg/ml, NIC=0.461. B Is electron cloud density/effective atomic number diagram in arterial phase, Zeff=8.82, Mix-0.6=100.2. C Is a 40keV single-energy diagram in the arterial phase. D Is the ROI spectrum curve in the arterial phase, λHU=4.79. E Is the diagram of iodine-based substances in the venous phase, IC=2.3mg/ml, NIC=0.44. F Is the diagram of electron cloud density/effective atomic number in the venous phase, Zeff=8.84, Mix-0.6=101.3. G Is the single-energy diagram at 40keV, in the venous phase. H Is the ROI spectrum curve in the venous phase, λHU=4.71

All image analyses and parameter measurements were independently performed by two senior radiologists with over 10 years of experience. The intraclass correlation coefficient (ICC) was calculated for each parameter to assess inter-observer agreement.

Efficacy evaluation

The imaging regression of lymph nodes one year after radiotherapy is used as the criterion for distinguishing between benign and malignant cases. Follow-up evaluations used the same models of MRI and CT equipment as before treatment to minimize measurement bias. Lymph node response was evaluated according to the self-defined criteria of this study (referring to RECIST1.1 [22] but adapted to small nodes), complete response (CR) was defined as disappearance of the small node; partial response (PR) as a ≥ 30% reduction in short-axis diameter; stable disease (SD) as a reduction < 30% or increase < 20%; and progressive disease (PD) as an increase ≥ 20%. Nodes were then categorized into the effective disease group (ED, CR + PR) and the stable group (SD). Based on their anatomical location, cervical lymph nodes were divided into three groups using the hyoid body and the lower edge of the cricoid cartilage as boundaries:

• Upper cervical lymph nodes: lymph nodes in the retropharyngeal region, level I, and level II.

• Middle cervical lymph nodes: lymph nodes in the upper parts of levels III, V, and VI.

• Lower cervical lymph nodes: lymph nodes in level IV, the supraclavicular fossa, and levels V and VI.

Statistical analysis

SPSS statistical software (version 26.0; IBM Corp., Armonk, NY, USA) and GraphPad Prism 8.3.0 (GraphPad Software, San Diego, CA, USA) were used. The Kolmogorov-Smirnov test was used to evaluate the normality of quantitative data. Normally distributed measurement data are expressed as mean ± standard deviation, whereas the non-normally distributed measurement data are expressed as median. The independent samples t test, Pearson χ² test, and Mann-Whitney U test were used to compare the differences between the two groups. A receiver operating characteristic (ROC) curve was drawn to evaluate the prediction performance of the binary logistic regression model. The best cutoff value was determined using the maximum Jorden index, and the area under the curve (AUC), sensitivity, and specificity under the ROC curve were calculated. The Delong test was used for pairwise comparisons of ROC curves. P < 0.05 was considered statistically significant.

Results

Participants and lymph node characteristics

The study included 125 small lymph nodes from 98 patients with NPC. After radiotherapy, 65 lymph nodes were assigned to the ED group, and 60 lymph nodes were assigned to the SD group. The median lymph node size was 6 mm (range, 5–9 mm). No extracapsular invasion or necrosis was found. Small lymph nodes were more common in stage III-IV NPC, mainly located in region II. Small lymph nodes in the lower neck were present in all patients in the ED group and were in complete response. Among the evaluated factors, N stage, lymph node levels, and short-axis diameter showed statistically significant differences between the two groups (Table 1).

Table 1.

Characteristics of small lymph nodes in nasopharyngeal carcinoma

Characteristics	ED group[n(%)]	SD group[n(%)]	P value
Total	65(100)	60(100)
T stage*
T2	11(16.9)	9(15)	0.259
T3	18(30.8)	24(40)
T4	29(52.3)	27(45)
N stage*
N1	20(30.8)	23(38.3)	0.036
N2	13(20)	20(33.3)
N3	32(49.2)	17(28.4)
TNM stage*
III	22(33.8)	23(38.3)	0.790
IV	43(66.2)	37(61.7)
Histology type
II	6(9.2)	4(6.7)	0.598
III	59(90.8)	56(93.3)
pre-EBV DNA
Negative (< 400)	31(47.7)	35(58.3)	0.234
Positive (> 400)	34(52.3)	25(41.7)
Lymph node levels
I	5(7.7)	8(13.3)	<0.001
II	33(50.8)	49(81.7)
III	13(20)	3(5)
IV	9(13.8)	0(0)
V	5(7.7)	0(0)
Short diameter (mm)
5–6	21(38.5)	48(70)	<0.001
7–9	37(61.5)	19(30)

^*T, N, and TNM stages were assessed by the 8th edition of the American Joint Committee on Cancer Union for International Cancer Control stage classification system

Abbreviations: ED effective disease, SD stable disease

DECT parameter comparison for small lymph nodes

The median ICC values for all quantitative parameters measured by the two radiologists were 0.901 (range 0.821–0.939) for the arterial phase and 0.928 (range 0.891–0.956) for the venous phase, and the parameters followed a normal distribution. NIC, IC, Mix-0.6, λHU, and Zeff were higher in the ED group than those in the SD group in both arterial and venous phases. Arterial NIC, IC, Mix-0.6, and λHU and venous NIC, IC, Mix-0.6, and λHU showed statistically significant differences between the two groups (P < 0.05) (Table 2, Fig. 2).

Table 2.

DECT parameter characteristics of small lymph nodes in nasopharyngeal carcinoma

DECT parameters	ED group	SD group	P value
Arterial phase
NIC	0.17 ± 0.07	0.15 ± 0.03	<0.001
IC(mg/mL)	2.06 ± 0.92	1.84 ± 0.41	<0.001
Mix-0.6	82.06 ± 20.96	81.63 ± 9.87	0.012
λHU(HU/keV)	3.79 ± 1.31	2.69 ± 0.51	<0.001
Zeff	8.57 ± 0.39	8.56 ± 0.2	0.123
Venous phase
NIC	0.44 ± 0.09	0.42 ± 0.09	0.045
IC(mg/mL)	2.52 ± 0.56	2.41 ± 0.61	0.022
Mix-0.6	90.62 ± 14.33	88.88 ± 13.78	0.024
λHU(HU/keV)	4.64 ± 1.05	3.51 ± 0.05	<0.001
Zeff	8.8 ± 0.25	8.76 ± 0.28	0.083

Abbreviations: DECT dual-energy computed tomography, ED effective disease, SD stable disease, IC iodine concentration, NIC normalized iodine concentration, Mix-0.6 mixing ratio of 0.6, λHU slope of the energy spectrum, Z_eff effective atomic number

Fig. 2.

Comparison of arterial and venous parameters of small lymph nodes between ED and SD groups: A NIC, B IC, C Mix-0.6, D λHU, and E Z_eff. Note: *P < 0.05, ***P < 0.001. Abbreviations: ED, effective disease; SD, stable disease; AP, arterial phase; VP, venous phase; IC, iodine concentration; NIC, normalized iodine concentration; Mix-0.6, mixing ratio of 0.6; λHU, slope of the energy spectrum; Z_eff, effective atomic number

Evaluating the diagnostic efficacy of DECT parameters of small lymph nodes

Binary logistic regression analysis showed that arterial NIC and λHU, and venous NIC, Mix-0.6, and λHU were independent risk factors for differentiating the sensitivity of small lymph nodes to chemoradiotherapy (P < 0.05). The ROC curve analysis showed that the AUC values for DECT parameters in the arterial and venous phases were 0.61–0.73 and 0.69–0.80, respectively, indicating that the diagnostic efficacy of venous phase parameters was higher. The single parameter λHU showed the highest diagnostic performance, with an AUC of 0.803, an optimal cutoff value of 4.465, and a Youden index of 0.503. The corresponding sensitivity and specificity values were 72.4% and 80.8%, respectively. When the AUC value of λHU increased to 0.91 with short diameter, lymph node partitioning, and the venous phase, the sensitivity and specificity were 92% and 85%, respectively (Tables 3 and 4, Fig. 3).

Table 3.

Multivariate analysis of DECT parameters in small lymph nodes of nasopharyngeal carcinoma

DECT parameters	β	S.E.	Wald	OR	95% CI	P value
Arterial phase
NIC	-0.082	0.044	3.413	0.922	0.845–1.005	0.045
λHU	0.260	0.102	6.454	1.296	1.061–1.584	0.011
Venous phase
NIC	0.085	0.037	5.250	1.089	1.012–1.171	0.022
Mix-0.6	-0.100	0.052	3.744	0.905	0.82–1.01	0.050
λHU	0.196	0.084	5.495	1.216	1.033–1.433	0.019

Abbreviations: DECT dual-energy computed tomography, S.E. standard error, OR odds ratio, CI confidence interval, NIC normalized iodine concentration, λHU slope of the energy spectrum, Mix-0.6 mixing ratio of 0.6

Table 4.

Diagnostic efficacy of DECT parameters in small lymph nodes of nasopharyngeal carcinoma

DECT parameters	AUC	95%CI	Youden	Cutoff	sensitivity(%)	specificity(%)
Arterial phase
NIC	0.655	0.566–0.745	0.338	0.163	56.9	76.9
IC(mg/mL)	0.649	0.562–0.736	0.246	1.850	62.1	62.5
Mix-0.6	0.614	0.521–0.707	0.281	78.200	48.3	79.8
λHU	0.726	0.647–0.805	0.372	3.475	77.6	72.2
Zeff	0.674	0.587–0.76	0.348	8.455	48.3	83.7
Venous phase
NIC	0.769	0.691–0.847	0.484	0.437	63.8	84.6
IC(mg/mL)	0.719	0.633–0.806	0.368	2.450	48.3	88.5
Mix-0.6	0.688	0.601–0.776	0.317	86.050	58.6	73.1
λHU	0.803	0.734–0.871	0.503	4.465	72.4	80.8
Zeff	0.735	0.654–0.817	0.397	8.695	81.0	58.7

Abbreviations: DECT dual-energy computed tomography, AUC area under the curve, CI confidence interval, IC iodine concentration, NIC normalized iodine concentration, Mix-0.6 mixing ratio of 0.6, λHU slope of the energy spectrum, Z_eff effective atomic number

Fig. 3.

ROC curves of DECT parameters of small lymph nodes in arterial phase (A), venous phase (B) and combined (C). Abbreviations: AUC, area under the curve; AP, arterial phase; VP, venous phase; IC, iodine concentration; NIC, normalized iodine concentration; Mix, mixing ratio; λHU, slope of the energy spectrum; Z_eff, effective atomic number; SD, short-axis diameter; Div, lymph node location in the lower neck

Discussion

An accurate assessment of the presence of metastatic cervical lymph nodes is the most important factor influencing prognosis and treatment planning for NPC. Patients with clinically diagnosed N0 disease have a high probability (≥ 20%) of occult neck lymph node metastasis [23]. Pathological biopsy is the gold standard for nodal metastasis, but radiotherapy is the primary treatment for NPC, making invasive biopsy impractical for all small lymph nodes. MRI uses a short-axis diameter > 10 mm as a diagnostic criterion, yet substantial inter-individual variation in nodal size precludes a universally optimal threshold [24, 25]. Small lymph nodes < 10 mm are common in NPC patients, and distinguishing inflammatory hyperplasia from occult metastasis is challenging in clinical practice. Studies have suggested that the number of suspicious micro-nodes correlates positively with N stage, and patients with bulky metastatic nodes often have a higher number of concurrent suspicious small nodes [10]. In addition to the MRI diagnostic criteria for post-pharyngeal and cervical lymph node metastases in the international consensus guidelines [11], it is also recommend combining follow-up MRI findings: nodes showing significant shrinkage (≥ 30%) compared with pre-treatment images, or nodes that remain stable immediately after radiotherapy but show progression on subsequent follow-up MRI, are often considered metastatic [10, 25].

DECT can generate images for a variety of substances (primarily water, iodine, and calcium) that can be quantitatively analyzed. This process creates tissue characteristic maps that reflect histochemical components and quantify their concentrations [26]. Numerous studies have demonstrated that these DECT quantitative parameters can characterize tumor blood supply and reflect lymph node features [27–30]. And DECT can provide quantitative iodine-based and energy spectrum parameters, objectively reflect the blood supply and tissue composition of lymph nodes, and has additional identification value for MRI-negative small lymph nodes. Moreover, its fast scanning speed and high clinical accessibility make it suitable as a routine pre-treatment assessment in NPC.

This study focuses on MRI-negative small lymph nodes and confirms that DECT can provide additional diagnostic information, partially compensating for the lack of MRI in the evaluation of micro-metastasis. Our study also found that patients with high T and N stages of NPC had more small lymph nodes, more small lymph nodes in region II, and SD accounted for the majority of cases. Notably, all lower neck small lymph nodes disappeared after radiotherapy, suggesting a higher metastatic risk in that region. If we assume that these completely regressed lower neck nodes were metastatic, then 4 patients would have undergone N stage reclassification: 2 from N1 to N2 (2.1%) and 2 from N2 to N3 (2.1%).

Chen et al. [12] studied the diagnostic value of shear wave elastography in judging small lymph node metastasis of NPC. More than 90% of small positive lymph nodes were detected in T3–T4 NPC, and 10% of 62 patients had an N stage change, including five cases from N1 to N2 and 1 case from N2 to N3. Using PET/CT to examine lymph nodes in NPC, Peng et al. [31] found that for lymph nodes measuring < 7 mm, the metastasis rate were very low. The metastasis rate of 7–9-mm lymph nodes was significantly increased in the I–IV/Vb regions, at 84.6%. The metastasis rate in lymph nodes ≥ 9 mm in the III/Va and IV/Vb regions was ≥ 90%, suggesting that small lymph nodes in the lower neck were more likely to harbor metastases. This result is consistent with our findings.

Our study found, for the first time, that λHU in the venous phase was the single parameter with the highest diagnostic performance for differentiating the sensitivity of small lymph nodes to chemoradiotherapy in NPC (AUC = 0.8). The results of this study were consistent with those of Lin et al. in identifying metastatic lymph nodes from colon cancer and Yu et al. in identifying metastatic lymph nodes from branchial adenocarcinoma (AUC = 0.8) [32, 33].

Combining imaging and pathophysiological mechanisms, the superior diagnostic performance of venous-phase λHU may be attributed to: (1) the lymph nodes in the ED group, active tumor cell proliferation is accompanied by abundant neovascularization and impaired endothelial integrity, leading to more pronounced iodine contrast agent retention and extravasation during the venous phase, resulting in a distinct spectral attenuation pattern compared with the lymph nodes in the SD group; (2) venous-phase perfusion is more stable and less affected by transient arterial flow fluctuations, thereby improving the measurement reliability of λHU.

Combining λHU in the venous phase with short-axis diameter and lymph node partition further improved the diagnostic power. When combined with the short-axis diameter of 7–9 mm, lymph node location in the lower neck, and a λHU of > 4.46 in the venous phase, the combined diagnostic power reaches 0.87, providing higher value for differentiating the sensitivity of small lymph nodes to chemoradiotherapy in NPC.

Few studies have examined the number and prognosis of small lymph nodes in NPC. Two retrospective studies found that small lymph nodes did not affect the prognosis of patients with NPC; however, neither study provided specific irradiation doses for small lymph nodes [34]. The management of small lymph nodes was largely based on the experience of individual clinicians’ experience. Large datasets suggest that the neck lymph node status is a major factor affecting survival in patients with NPC [35]. Our integrated model combining short-axis diameter, nodal region, and DECT parameters may improve diagnostic efficiency for the sensitivity to chemoradiotherapy and provide objective evidence for clinical decision-making regarding small lymph nodes.

Due to the constraints of NPC diagnosis and anatomical limitations, it was not feasible to perform puncture or surgical pathology on all small nodes; we therefore used radiological regression after radiotherapy as a surrogate reference. Regression of small lymph nodes after chemoradiotherapy could result from tumor regression or from resolution of local inflammation; moreover, inter-patient and inter-lesion heterogeneity in chemoradiosensitivity may introduce classification bias. Accordingly, the diagnostic performance reported here reflects the correlation between DECT parameters and post-treatment regression, which does not fully equate to pathological confirmation of metastasis. Other limitations of this study include: (1) this is a single-center retrospective study with a limited sample size and no external validation cohort, which restricts generalizability of the findings; (2) all imaging data were acquired on the same Siemens SOMATOM Definition Flash DECT scanner. The optimal venous-phase λHU cutoff (> 4.46) derived herein is specific to this scanner and its acquisition parameters and cannot be directly applied to other DECT brands or models. Future cross-device calibration and multicenter external validation are needed; (3) small lymph nodes are susceptible to partial volume effects. Despite our standardized ROI protocol, minor interference with measurement stability cannot be excluded; (4) this is a hypothesis-generating study, and the results need to be further confirmed by large-sample, multi-center, and external validation studies.

Conclusion

Small cervical lymph nodes deserve attention during IMRT for NPC, especially those in the lower neck, which have a higher rate of complete response after radiotherapy. This study is the first to apply DECT quantitative parameters to demonstrate significant differences between small lymph nodes with differing chemoradiosensitivity. Venous-phase λHU is the single parameter with the highest diagnostic performance. A model combining short-axis diameter and nodal region further improves the differentiation efficiency for chemoradiosensitivity and may assist in clinical assessment of metastatic risk. Future multicenter, large-sample studies with selective invasive pathological confirmation are warranted to validate our findings.

Authors’ contributions

ZL contributed to the study conceptualization. Methodology, data curation, formal analysis were performed by HH, DL and HZ. The original draft of the manuscript was written by HD and ZL. English language editing were provided by JS. The review and editing of the manuscript was performed by GH and MK. All authors read and approved the final manuscript.

Funding

This work was supported by grants from the National Natural Science Foundation of China (No. 82272736, 82160217, 82160467, 81460460, 81760542, 82560224); Guangxi Science Fund for Distinguished Young Scholars (No.2024JJG140004); the Joint Project on Regional High-Incidence Diseases Research of the Guangxi Natural Science Foundation (No. 2023GXNSFBA026211, 2024GXNSFAA010026); Innovation Project of Guangxi Graduate Education (No. YCSW2024254, YCSW2025279, YCBZ2024125); the Research Foundation of the Science and Technology Department of Guangxi Province, China (No. 2023GXNSFDA026009, 2016GXNSFAA380252, 2018AB61001); the Research Foundation of the Health Department of Guangxi Province, China (No.S2018087); Guangxi Medical High-level Talents Training Program; the Central Government Guide Local Science and Technology Development Projects (No. ZY18057006, ZY23055031); the Guangxi Scholarship Fund of Guangxi Education Department of China and Key Laboratory of Early Prevention and Treatment for Regional High Frequency Tumors (Guangxi Medical University), and the Ministry of Education (No. GKE-ZZ202507, GKE-ZZ202302).

Data availability

The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the Helsinki Declaration and approved by the Ethics Committee of the First Affiliated Hospital of Guangxi Medical University (Approval Number: 2023-E329-01). All patients provided informed consent and their information was confidential.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.