Surgical outcomes of primary pediatric spinal cord tumors: A retrospective cohort

Abstract

Background:

Pediatric primary spinal cord tumors are rare, accounting for approximately 6% of central nervous system neoplasms. Most published data come from high-income countries (HICs) with access to early diagnosis, intraoperative monitoring, and specialized neuro-oncology units. In low- and middle-income countries (LMICs), diagnostic and therapeutic resources are limited. This report summarizes the first cohort of pediatric cases from Pakistan, highlighting surgical outcomes and systemic challenges in a resource-constrained setting.

Methods:

We conducted a retrospective cohort study of patients ≤18 years who underwent surgery for primary pediatric spinal tumors at a tertiary hospital in Karachi (2008–2025). Demographics, presentation, tumor location and compartment, histopathology, extent of resection, postoperative complications, adjuvant therapy status, mortality, recurrence/progression, and follow-up outcomes were extracted from records.

Results:

Nineteen children were included (11 female; median age, 12 years). Lower-limb weakness (n = 11) and back pain (n = 7/15 with available data) were the most common presentations, with a median diagnostic delay of 90 days. Tumors were most often lumbar (n = 6), cervical (n = 5), or thoracic (n = 5). Among 13 patients with documented compartment data, 7 tumors were intramedullary, 4 were extramedullary/intradural or paramedullary, and 2 were intradural without further compartment specification. Ependymoma (n = 5) and astrocytoma (n = 4) predominated. Gross total resection (GTR) was documented in 10 patients and maximal safe/subtotal resection (MSR/STR) in 7. Postoperative complications were documented in 2/16 patients with available data. At the last follow-up, 5/12 patients with neurological follow-up improved, 5 remained stable, and 2 worsened. Mortality was documented in 5/17 patients with survival status available patient had recurrence after GTR, and 1 had progression after MSR/STR.

Conclusion:

Surgical management of pediatric primary spinal tumors in an LMIC tertiary center was feasible, with limited documented perioperative morbidity after data correction. However, small sample size, missing event-level data, and a short median follow-up of 4 months limit interpretation of recurrence, progression-free status, and mortality. Earlier diagnosis, standardized surveillance, multidisciplinary review, rehabilitation, infection prevention, and improved access to adjuvant therapy may help strengthen care pathways in resource-constrained settings.

INTRODUCTION

Primary pediatric spinal cord tumors are neoplasms arising within the spinal cord or its immediate coverings in children.[19] Central nervous system (CNS) tumors are the second most common childhood cancer,[16] but spinal cord tumors account for only a small fraction (around 1–10%) of CNS tumors in this age group.[6,19] For example, large series reports that only a few percent of spinal tumor surgeries involve pediatric patients.[13] Spinal tumors often cause severe neurologic deficits: patients frequently present with pain, limb weakness, sensory loss, or bladder/bowel dysfunction[19] and disease progression can result in permanent paralysis.[11] Because surgical resection is the mainstay of treatment,[19] maximizing neurologic preservation is critical. These neurologic impairments occur during a child’s development and can profoundly disrupt physical growth, mobility, and quality of life. Characterizing these surgical and functional outcomes is therefore essential for guiding care.

Existing literature, largely from high-income centers, indicates that factors such as tumor histology and preoperative neurologic status influence outcomes.[8,19] However, reported outcomes vary widely across series, and prognostication remains challenging.[11] Importantly, most data come from well-resourced settings, and there is a dearth of outcome data from low- and middle-income countries (LMICs).[15] For example, childhood brain tumor survival rates in many LMICs remain below 60%, illustrating broader disparities.[7] In Pakistan specifically, no prior series of pediatric spinal cord tumors has been reported, looking at patient outcomes for children diagnosed with primary spinal cord tumors (PSCTs), emphasizing the need for regional outcome data.

To address these gaps, we conducted a retrospective cohort study of children with PSCTs treated at our tertiary academic center in Karachi. The objectives were to characterize tumor histology, tumor compartment, surgical management, postoperative morbidity, follow-up neurological outcomes, recurrence/progression, and survival, and to explore associations with neurological status change at last follow-up. We hypothesized that more extensive resection and better preoperative neurologic status would be associated with superior functional recovery.

MATERIALS AND METHODS

This retrospective cohort study included patients diagnosed with pediatric primary spinal tumors and treated at a tertiary care hospital between 2008 and 2025. The study was conducted in accordance with the STROBE guidelines for observational studies as mentioned in Supplementary File 1. Inclusion criteria comprised patients ≤18 years with histologically confirmed primary spinal tumors involving the spinal cord, intradural compartment, or immediately adjacent paramedullary/filum region who underwent surgical intervention during the study period. Exclusion criteria included incomplete clinical data or loss to follow-up before initial postoperative assessment. Ethical approval was obtained from the Institutional Review Board (ERC# 2023-8992-26425), and informed consent was waived according to institutional guidelines.

Data collection and variables

Demographic data collected included age at diagnosis, sex, and duration of symptoms before presentation. Clinical characteristics included presenting symptoms, tumor location along the spinal axis (cervical, thoracic, lumbar, cervicothoracic, or thoracolumbar), tumor compartment (intramedullary, extramedullary/intradural or paramedullary, or intradural not otherwise specified when source records did not distinguish compartment), histological subtype, and tumor grade according to the World Health Organization (WHO) classification. Surgical variables included extent of resection, categorized as gross total resection (GTR), maximal safe/subtotal resection (MSR/STR), partial resection, or biopsy where documented; operative time; and perioperative complications. Adjuvant radiotherapy, chemotherapy, or other postoperative oncologic treatment was also considered for extraction. Neurological outcomes were assessed using the modified McCormick scale (MMS) at baseline, immediate postoperative assessment, and last follow-up. Recurrence/progression status and survival data were collected when available. All variables were independently extracted from electronic medical records and operative reports.

Outcome measures

The primary outcome was neurological improvement at last follow-up, defined as a decrease in MMS grade relative to baseline. Secondary outcomes included postoperative complications, tumor recurrence after GTR, tumor progression after less-than-total resection (MSR/STR, partial resection, or biopsy), length of hospital stay, disease status at follow-up, adjuvant therapy, and overall survival. Disease status was reported descriptively at the last available assessment; no time-to-progression or progression-free survival analysis was performed. Recurrence was defined as new or recurrent tumor after documented GTR. Progression was defined as enlargement or worsening of residual tumor after less-than-total resection. Postoperative complications were defined as any wound or systemic event occurring within the perioperative period. Survival was defined from the date of surgery to the date of death or last follow-up.

Statistical analysis

Continuous variables were summarized as medians with interquartile ranges (Q1–Q3), and categorical variables were summarized primarily as counts because of the small sample size. Percentages were used sparingly only when helpful and when denominators were clear. Pairwise Wilcoxon signed-rank tests were used to compare MMS grades at different time points. Associations between demographic, tumor, and surgical factors and neurological improvement were assessed using Fisher’s exact or Chi-square tests for categorical variables and Kruskal–Wallis or Wilcoxon rank-sum tests for continuous variables. Exact logistic regression was performed to identify predictors of neurological improvement, with odds ratios (OR) and 95% confidence intervals (CIs) reported. Kaplan–Meier survival curves were generated for patients with available event-time data and stratified descriptively by extent of resection (GTR vs. MSR/STR), with differences assessed using the log-rank test. Analyses were performed using STATA v17 software, and statistical significance was set at a two-sided P < 0.05.

Follow-up

Patients were followed from the date of surgery until the last clinical assessment or death, with follow-up duration reported in months. Median follow-up time was 4 months. This follow-up interval is short for pediatric spinal tumors and substantially limits interpretation of recurrence, progression-free status, and mortality. Survival analyses accounted for censored observations due to incomplete follow-up when event-time data were available. Missing data were not imputed; analyses were performed using available cases only, and denominators are reported for each outcome.

RESULTS

Demographic and clinical characteristics

The study included 19 patients with primary pediatric spinal tumors, with a median age of 12 years. Females comprised slightly more than half of the cohort (n = 11). The most common presenting symptom was lower-limb weakness (n = 11), followed by back pain (n = 7/15 with available data), and the median duration of symptoms before diagnosis was approximately 3 months. Tumors were distributed across the spinal axis, most frequently in the lumbar (n = 6), cervical (n = 5), and thoracic (n = 5) regions. Compartment data were available for 13 patients: 7 tumors were intramedullary, 4 were extramedullary/intradural or paramedullary, and 2 were intradural without further specification; 6 cases lacked documented compartment data. Ependymoma (n = 5) and astrocytoma (n = 4) were the most common histological types, and low-grade tumors (WHO Grade I-II; n = 14) predominated. Patient characteristics are summarized in Table 1.

Table 1:

Demographic and clinical characteristics of patients with spinal tumors (n=19).

Surgical, postoperative, and follow-up outcomes

The extent of resection was documented as GTR in 10 patients and MSR/STR in 7 patients. The remaining 2 cases could not be classified from the submitted dataset. Postoperative complications were documented in 2/16 patients with available complication data, both of which were infectious in nature. Tumor recurrence after GTR occurred in 1/10 patients, and tumor progression after MSR/STR occurred in 1/7 patients. These outcomes are summarized in Table 2.

Table 2:

Surgical, postoperative, and follow-up outcomes of patients with spinal tumors.

Immediately postoperatively, 4/12 patients with available neurological status demonstrated improvement, 8/12 were stable, and no immediate postoperative neurological deterioration was documented. At the last follow-up, among 12 patients with available MMS data, 5 improved, 5 remained stable, and 2 worsened. Neurological status at last follow-up varied across the MMS grades, reflecting persistent functional deficits in a subset of patients.

Mortality at last follow-up was documented in 5/17 patients with available survival status. Mortality was analyzed separately from disease status and should not be interpreted as the inverse of progression-free status. Among 12 patients with disease-status follow-up, 10 had no documented recurrence/progression or were clinically stable/improved, whereas 2 had recurrent or progressive disease (1 recurrence after GTR and 1 progression after MSR/STR). No statistically significant associations were observed between neurological improvement at last follow-up and age, gender, duration of symptoms, tumor location, histology, tumor grade, extent of resection, postoperative complications, operating time, or immediate postoperative MMS score (all P > 0.05) as demonstrated in Table 3.

Table 3:

Association between patient, tumor, and surgical factors and neurological improvement at last follow-up.

Patients without documented postoperative complications had mixed neurological outcomes, and the single patient with a documented complication in the follow-up analytic cohort remained neurologically stable. Overall, neurological recovery appeared multifactorial and could not be attributed to any single demographic, tumor, or surgical factor in this small cohort.

In exact logistic regression analysis, neither extent of resection (GTR vs. non-GTR) nor immediate postoperative neurological status (good vs. not good) was significantly associated with neurological improvement at last follow-up (OR 0.59, 95% CI 0.01–14.56, P = 1.000; OR 0.56, 95% CI 0.08–2.28, P = 0.571, respectively), as shown in Table 4.

Table 4:

Predictors of neurological improvement at last follow-up: Exact logistic regression analysis.

Median MMS at different time periods

As illustrated in Figure 1, the median MMS grade was 3 at diagnosis and 3 immediately postoperatively, with a modest decrease to 2.5 at last follow-up. Because lower MMS grades indicate better neurological function, these descriptive medians are compatible with modest improvement by follow-up. Because lower MMS grades indicate better neurological function, these descriptive medians are compatible with modest improvement by follow-up. Because the medians are based on available observations at each time point, they should not be interpreted as a paired longitudinal estimate.

Figure 1:

Line graph of observed median Modified McCormick Scale (MMS) grades among patients with data available at each time point. Lower MMS grades indicate better neurological function.

Pairwise Wilcoxon signed-rank tests were performed to assess changes in MMS grades over time. No comparison reached statistical significance: preoperative versus immediate postoperative MMS (exact P = 0.125), immediate postoperative versus last follow-up MMS (exact P = 0.7266), and preoperative versus last follow-up MMS (exact P = 0.2188). These findings should be interpreted cautiously because only patients with available paired MMS data were included.

Survival analysis

Overall survival status was available for 17 patients, among whom 5 deaths were documented. Kaplan–Meier survival analysis was restricted to 11 patients with available event-time data; 5 deaths were observed in this analytic subset. Cause and timing of death were not available for any of the patients.

As shown in Figure 2, overall survival in the analyzable Kaplan–Meier cohort declined descriptively over follow-up; median survival was not reached because fewer than half of the analyzable patients experienced the event. Given missing event-time data and short follow-up, the Kaplan–Meier estimates should be considered exploratory.

Figure 2:

Kaplan–Meier survival estimate for patients with analyzable event-time data (n = 11). The y axis is survival probability.

Patients were stratified by extent of resection for exploratory Kaplan–Meier analysis: GTR (n = 7 with analyzable event-time data) versus MSR/STR (n = 4 with analyzable event-time data), as shown in Figure 3. The log-rank comparison was not statistically significant (P = 0.36).

Figure 3:

Kaplan–Meier survival estimates stratified by extent of resection: Gross total resection (GTR) (n = 7) versus Maximal safe/subtotal resection (MSR/STR) (n = 4) among patients with analyzable event-time data. The y axis is survival probability.

The MSR/STR subgroup contained only 4 analyzable patients; therefore, the apparent difference between curves should not be interpreted as evidence that the extent of resection independently determined survival. Resection extent is likely confounded by tumor compartment, histology, infiltrative biology, preoperative neurological status, and surgical risk.

Overall, survival analyses are presented descriptively and require confirmation in a larger cohort with complete event-time data and longer follow-up.

Kaplan–Meier curve showing overall survival stratified by extent of resection. The difference between GTR and MSR/STR groups was not statistically significant (log-rank test: Chi-square (1) = 0.84, P = 0.36), and the small subgroup sizes preclude causal inference.

DISCUSSION

In this single-center cohort of pediatric PSCT, patients were predominantly early adolescents with a median age of 12 years, consistent with prior literature.[11] The most common presenting symptom was lower limb weakness, followed by back pain, indicating progressive mass effect of intramedullary and extramedullary lesions within the spinal canal.[3] The cohort included both intramedullary tumors and extramedullary/intradural or paramedullary lesions; therefore, outcome interpretation should account for tumor compartment rather than treating all lesions as biologically equivalent. Most of the cases had a subacute clinical course, with a median duration of symptoms before diagnosis lasting 3 months. This is shorter than most studies which describe delays in diagnosis up to 12 or even 36 months in some cases, depending on the type and grade of the tumors.[4,5]

Tumor distribution across the spine was relatively balanced, with lumbar lesions slightly predominating. Among patients with documented compartment data, intramedullary tumors were the largest subgroup, followed by paramedullary and then intradural. The predominance of glial tumors, particularly ependymoma and astrocytoma, is concordant with established epidemiology.[4,19] Low-grade tumors (WHO Grade I-II) represented the dominant subgroup, consistent with prior pediatric series.[19]

Despite the predominance of low-grade histology, functional outcomes were heterogeneous. Immediate postoperative neurological improvement occurred in 4/12 patients with available data, and no immediate postoperative neurological deterioration was documented. At the last follow-up, 5/12 patients improved, 5 remained stable, and 2 worsened. The absence of statistically significant paired MMS changes suggests that the observed improvements should be interpreted cautiously, particularly given missing follow-up data and the short median follow-up interval. This trajectory aligns with current literature which indicates that early decompression may reverse reversible cord dysfunction, with long-standing deficits indicating irreversible myelopathic injury.[18]

The extent of resection was not significantly associated with neurological improvement at the last follow-up. Importantly, recurrence and progression were analyzed as distinct outcomes: recurrence was assessed after GTR, whereas progression was assessed after less-than-total resection. After correction of the yes/no coding, recurrence after GTR was documented in 1/10 patients and progression after MSR/STR in 1/7 patients. These small counts do not support an inference that GTR was associated with high recurrence or that MSR/STR was associated with frequent progression. Instead, they emphasize the need for careful postoperative surveillance and longer follow-up.

Mortality was documented in 5/17 patients with available survival status; however, Kaplan-Meier analysis included only 11 patients with event-time data. Causes and timing of death were unavailable. Documented postoperative complications were uncommon after correction (2/16), and available data therefore do not support a claim that perioperative morbidity drove mortality in this cohort.

None of the examined demographic, tumor-related, or surgical variables was significantly associated with neurological improvement at last follow-up. This likely reflects limited statistical power, sparse outcome events, incomplete follow-up, and biological heterogeneity across intramedullary and extramedullary/intradural lesions. Current WHO CNS classification increasingly emphasizes molecular features that influence biological behavior and recurrence risk.[12,17] The absence of routine molecular profiling in our cohort limits risk stratification.

Importantly, none of the examined demographic, tumor-related, or surgical variables were significantly associated with neurological improvements at the last follow-up. While this reflects multifactorial determinants of outcome, it is more likely attributable to limited statistical power and less events. Regardless, rare diseases such as pediatric PSCTs are inherently challenging to study in single-center cohorts.

When interpreting outcomes in this LMIC cohort, several contextual challenges are relevant. The manuscript demonstrates delayed presentation, incomplete follow-up, and missing event-level data; these findings are consistent with broader barriers in resource-constrained settings, including referral delays, cost constraints, variable access to timely magnetic resonance imaging (MRI), difficulty maintaining postoperative surveillance, limited rehabilitation access, and challenges obtaining adjuvant treatment. These mechanisms were not directly measured at the patient level in this study and should therefore be regarded as plausible system-level contributors rather than proven causal factors.[2,14] These limitations arise due to deficiencies in healthcare infrastructure, workforce shortages, and financial barriers that delay access to diagnosis and treatment. While the association of these factors has been studied extensively in pediatric patients with brain tumors, the effect of these factors in pediatric PSCTs remains sparse and under-reported.[9]

High-income country series and registry analyses generally report more mature follow-up, multidisciplinary treatment pathways, and 5-year overall survival that can exceed 90% in selected pediatric intramedullary spinal tumor cohorts.[1,10] The documented mortality proportion should be interpreted cautiously and should not be directly compared with 5-year survival estimates from selected external cohorts, given the short follow-up and incomplete event-time data. Potential strategies to improve outcomes in similar settings include earlier recognition of pediatric myelopathy, improved MRI access, standardized postoperative MRI surveillance, multidisciplinary tumor board review, rehabilitation integration, infection prevention, clear referral pathways, and improved access to indicated radiotherapy or chemotherapy. Such global health initiatives aimed at strengthening cancer surgical systems will serve to reduce these disparities and improve outcomes in resource-limited settings.[2]

Beyond its clinical findings, this study provides regional data from a tertiary care hospital in an LMIC, where pediatric primary spinal tumor outcomes remain scarce and underreported. Existing literature originates primarily from HICs with better neuro-oncology infrastructure, standardized adjuvant therapy protocols, and routine molecular classification. By characterizing tumor distribution, compartment, surgical management, neurological outcomes, and exploratory survival trends within a resource-limited setting, this study establishes the foundation for future multicenter collaborations. However, due to the short follow-up and incomplete data, these results should be used to generate hypotheses rather than to define definitive prognostic rules.

Limitations

This study is limited by its small sample, retrospective, single-center design, which introduces potential selection bias and limits generalizability. A median follow-up of only 4 months, which is very short for PSCTs, substantially limits interpretation of recurrence, progression-free status, late neurological recovery, and mortality. Follow-up and event-time data were incomplete, and causes of death, timing of recurrence/progression, adjuvant radiotherapy/chemotherapy status, and the exact extent of resection for 2 cases were not available due to being past medical reports. The lack of routine molecular diagnostics restricted classification according to contemporary WHO frameworks.[13] Finally, although MMS is widely used, it cannot adequately capture other aspects such as pediatric function, rehabilitation needs, school participation, or quality of life.

CONCLUSION

In this single-center LMIC cohort of pediatric primary spinal tumors, ependymoma and astrocytoma predominated, and most lesions were low grade at diagnosis. The cohort included both intramedullary and extramedullary/intradural or paramedullary tumors, which should be considered separately in future analyses. Documented perioperative complications (2/16), recurrence after GTR (1/10), and progression after MSR/STR (1/7) were uncommon among patients with available data. Mortality was observed in 5/17 patients with known survival status; cause-specific information was unavailable. Because follow-up was short and missing data were substantial, conclusions regarding disease status and prognostic factors must remain cautious. Future multicenter prospective studies with standardized imaging surveillance, molecular classification, adjuvant therapy documentation, rehabilitation outcomes, and longer follow-up are needed to refine prognostication and improve care.

Ethical approval:

The research/study was approved by the Institutional Review Board at Aga Khan University, number 2023-8992-26425, dated September 16, 2023.

Declaration of patient consent:

The authors certify that they have obtained all appropriate patient consent forms. In the form, the patient has given consent for clinical information to be reported in the journal. The patient understands that the patient’s names and initials will not be published and due efforts will be made to conceal their identity, but anonymity cannot be guaranteed.

Financial support and sponsorship:

Nil.

Conflicts of interest:

There are no conflicts of interest.

Use of artificial intelligence (AI)-assisted technology for manuscript preparation:

The authors confirm that there was no use of artificial intelligence (AI)-assisted technology for assisting in the writing or editing of the manuscript and no images were manipulated using AI.

Supplementary data available at:

https://doi.org/10.25259/SNI_325_2026

Disclaimer

The views and opinions expressed in this article are those of the authors and do not necessarily reflect the official policy or position of the Journal or its management. The information contained in this article should not be considered to be medical advice; patients should consult their own physicians for advice as to their specific medical needs.