Palpable craniocervical lymph nodes are a common finding in pediatric patients. These nodes are generally considered enlarged if measured more than 10 mm in their longest diameter (1). Enlarged nodes, i.e., lymphadenopathy, are arbitrarily categorized as acute, sub-acute and chronic, if persisting for <2, 2–6 and >6 weeks, respectively (2).

Lymphadenopathy may result from various etiologies, most of them infectious and inflammatory; malignancy is the culprit in a minority of cases. Although malignancy can be diagnosed by cytology of samples obtained using fine needle aspiration, the sensitivity of cytology in children is inadequate (3). Therefore, a definite diagnosis is based on a nodal biopsy. The decision to biopsy an enlarged lymph node is fairly easy in patients with systemic manifestations, and no evidence of infectious or inflammatory etiology. However, recommending this intervention in an otherwise healthy child with no abnormal blood tests or imaging studies is challenging given the potential surgical complications (nerve insults, infections, facial scar, the risks of a general anasthesia), costs of hospitalization and the high rates of futile pathology results that do not affect management (3–9). Finding pre-operative clinical tools that may assist physicians in the dilemma of whether to recommend surgical intervention can be of great benefit. The aim of the present study was to identify predictive factors for malignancy among generally healthy pediatric patients with focal subacute and chronic head and neck lymphadenopathy.

In this study of 450 children who were referred for evaluation of sub-acute or chronic focal craniocervical lymphadenopathy, multivariate analysis identified older age and larger nodal size as two factors predicting malignancy. The respective ORs were 1.649 (95%CI: 1.197–2.349, p = 0.004) for every increase in 3 years in age and 2.080 (95% CI: 1.292–3.330, p = 0.002) for every increase in nodal size of one centimeter. Moreover, we managed to quantify the risks for malignancy of specific age groups and nodal sizes. Accordingly, 20% of the patients older than 12.5 years and 12% of those with a lymph node diameter greater than 31 mm were diagnosed with malignancy.

The two factors identified herein were previously reported as associated with malignancy among children (5, 11–18). Soldes et al. identified an association between increasing age and the probability of malignancy (11), and others reported ORs of 1.072 (95%CI: 1.001–1.148) and 1.209 (95%CI: 1.082–1.350) for every one-year increase in age (12, 13). In a study that included children and adults, the mean age was significantly higher in the malignant than the non-malignant group (6). The increased incidence of lymphadenopathy in young children, which is probably due to a higher rate of pharyngeal infections in early childhood, can partially explain the lower rate of malignancy in younger than older children. Interestingly, a recent study from Korea found that malignancy was more common in younger than older children (5).

The results are inconsistent regarding the lymph node diameter cut-off values that would best imply likelihood of malignancy. However, most studies suggest that a nodal size above 2–3 cm is associated with an increased risk for malignancy. One study found that a cut-off of 2.6 cm yielded a sensitivity of 76.9% and a specificity of 67.4%, for malignancy, with a negative predictive value of 90.6%, and a positive predictive value of 41.7% (5). Others suggested that a diameter ≥2 cm yielded a higher likelihood of malignancy (15). Several pediatric studies reported statistically significant higher rates of a malignant disease compared to a benign condition in nodes with diameters >3 cm: 69.5% vs. 33.3%, 58.3% vs. 19% and 85.6% vs. 33.2% (16–18). Accordingly, a higher mean nodal diameter was reported in excised nodes with malignant compared to benign pathology results, 3.51 ± 2.78 vs. 2.61 ± 0.91 cm (12). In contrast, in a study of 98 children who underwent surgical excision of persistent enlarged lymph nodes, an association was not found between node size and a diagnosis of malignancy (19).

The rate of malignancy in our patients was substantially lower than previously reported: 5.5% vs. 23%–72% (5, 11, 16–18, 20). This disparity may be related to our cohort consisting of generally healthy children with sub-acute or chronic focal craniocervical lymphadenopathy as the only presumed risk factor for malignancy. In contrast, most other studies included diversified populations with characteristics that could affect the risk for malignancy. These included patients with a history of a neoplasia (11, 14), adults (14), patients with generalized lymphadenopathy (11, 13–18, 20) and children with acute presentation (17, 18, 21). Furthermore, some studies comprised only patients with post-nodal excisions, who were apparently a-priori at higher risk of malignancy (5, 11–14, 20). It is plausible that the presentation of relatively minor general manifestations among our patients with malignancy reflects successful containment of the proliferative process within the presented nodes. Indeed, in a recent study of asymptomatic pediatric patients with persistent cervical lymphadenopathy, the rate of malignancy was low, only 1.5% (4).

Similar to our study, Bozlac et al. looked for risk factors of malignancy among patients with cervical lymphadenopathy in a general pediatric clinic. In their prospective study, six patients were diagnosed with malignancy compared to 212 children with other conditions (21). This yielded a low rate of 2.8% of malignancy. In accordance with our results, the children with malignancies tended to be older than those without. However, unlike our findings, the association was not statistically significant, possibly due to the small number of patients with neoplasia. While 58.7% of their patients had acute symptoms, all our patients were evaluated for sub-acute or chronic lymphadenopathy. Another methodological difference is our longer follow-up period, 3.6 years (IQR: 2.1–4.9) vs. 8 weeks, which is crucial to avoid missing late diagnoses of malignancy.

The associations with malignancy found in our univariate analysis corroborate several other studies, specifically for parameters like weight loss (18, 22) supra-clavicular location (13, 14, 16–18, 22, 23), abnormal chest x-ray (13, 16–18) and targeted cervical sonography (4, 14, 20, 21). The diagnostic advantage of the latter has been supported by two findings. First, the maximal nodal width-to-length ratio aided in predicting malignancy, with an OR of 52.08 (95%CI: 16.10–168.6) (20). Second, the absence of a nodal fatty hilum was shown to be useful in deciding on surgical biopsy for suspected malignancy (4). However, Ingolfsdottir et al. reported an 80% false positive and 15% false negative result of pre-operative sonographic findings, compared to histology results of malignant and benign conditions, respectively (23). Others did not find statistically significant differences between sonographic results of patients with benign and malignant lymph nodes (22). These results underscore the caution needed in implying nodal sonographic interpretation in clinical decisions of lymphadenopathy evaluation.

In accordance with others, we did not find systemic symptoms, namely night sweats and fever, to be associated with malignancy (12, 22). We also did not find utility of erythrocyte sedimentation rate, CRP, complete blood count and hepatomegaly in predicting malignancy, thus corroborating other studies (6, 11, 13, 14, 16–18, 21). These results may be due to differences in tumor types and subtypes, and in the degree of their local and remote proliferation.

The shared decision-making model has been shown to decrease the number of biopsies in patients with lymphadenopathy (4, 6, 9). Employing this model by sharing in-house ENT and hematology-oncology physicians in the decision to recommend a biopsy probably contributed to the relatively low rate of invasive interventions in our cohort. It is plausible that this practice also contributed to parental confidence in the suggested management approach.

The strengths of our study are its large scale, the focus on a homogenous cohort with a specific condition, and the practical findings, which may be useful for clinicians evaluating pediatric patients with focal craniocervical lymphadenopathy. However, our study has several limitations. First is the retrospective design. Nevertheless, data were collected extensively from standardized electronic files, and included a wide range of demographic, clinical and laboratory variables, as well as imaging investigations. A second limitation is that some diagnoses of malignancy may have been missed, as not all the patients underwent a surgical biopsy and some may have been diagnosed after the study ended. However, this is unlikely due to the long duration of the follow-up i.e., the median time from the lymphadenopathy presentation to the last electronic search was 3.6 years (IQR: 2.1–4.9). Thirdly, our results represent a single-center experience of an ambulatory pediatric setting. Therefore, prospective studies of children with focal craniocervical lymphadenopathy are needed to validate our results, preferably with standardized indications for nodal biopsy.

In conclusion, our findings suggest that age and nodal size can contribute to clinicians’ decisions regarding excision of an enlarged focal craniocervical lymph node in a generally healthy child. A lymph node >2 cm in school-aged children or adolescents, or >3 cm in toddlers should be strongly considered for excision.