Delayed Diagnosis Linked to Poorer Treatment Response in Children With Brain Tumors

For children with central nervous system tumors, the clock between the first symptom and the confirmed diagnosis may matter more than clinicians have appreciated. A cohort study of 111 pediatric patients treated at a Mexican national pediatric institute found that when diagnosis took longer than one month, or longer than three months, from the first appearance of symptoms, the children were significantly more likely to fail to respond to treatment and significantly more likely to experience relapse. The findings, published in Frontiers in Pediatrics, come from one of the first studies to connect diagnostic delay directly to treatment outcomes in this population, rather than merely cataloguing the reasons that delays occur.

The research team reviewed electronic medical records of pediatric patients with newly diagnosed central nervous system tumors treated between October 2018 and October 2021, using the WHO 2016 classification for histopathological diagnosis. Nearly all of the children, 104 of 111, or 94 percent, had their diagnosis confirmed by histopathology from surgical resection or biopsy; the remaining seven, all with secreting germ cell tumors, were diagnosed through cerebrospinal fluid and serum tumor markers together with neuroimaging. Most children, 95 patients or 85.6 percent, underwent surgical resection as their first therapeutic step before adjuvant therapy, while 16 patients, or 14.4 percent, had a biopsy only followed immediately by chemotherapy or radiotherapy.

Boys made up 55 percent of the patients, and the median age at diagnosis was 12 years. High-grade tumors accounted for 64 percent of cases and low-grade tumors for 36 percent. Gliomas were the most common tumor type at 55 percent of the cohort, followed by embryonal tumors at 35 percent, germ cell tumors at 6 percent, and other tumor types at 4 percent. Disease progression was the predominant treatment response: at the end of the treatment protocol, 46 percent of children had progressive disease, 21.6 percent had stable disease, 12.6 percent had a partial response, and only 19.8 percent achieved a complete response.

The central measurement in the study was diagnostic delay, defined as the time between the onset of symptoms and the establishment of the tumor diagnosis, with the date of diagnosis fixed as the definitive event that confirmed the disease and authorized the start of oncological treatment. The median interval from symptom onset to diagnosis was 80 days, with an interquartile range of 29 to 195 days. The delays were strikingly common: 80 children, or 72.1 percent, waited more than one month for a diagnosis; 52 children, or 46.8 percent, waited more than three months; and 32 children, or 28.8 percent, waited more than six months.

To test whether these delays affected outcomes, the researchers used the Response Assessment in Neuro-Oncology, or RANO, criteria, standardized guidelines that combine tumor measurements with clinical status and leptomeningeal findings to evaluate how brain tumors respond to therapy. Children with complete or partial responses were classified as responders, while those with stable or progressive disease were classified as non-responders. Kaplan-Meier time-to-event analysis showed that a diagnostic delay of more than one month was significantly associated with a higher probability of non-response, with a p-value of 0.014, and delays of more than three months showed an even stronger association, with a p-value of 0.004. Delays of more than six months, however, did not reach statistical significance, with a p-value of 0.111.

The multivariable analysis strengthened the signal. In Cox proportional hazards regression adjusted for age and sex, a diagnostic delay of more than one month was associated with non-response to treatment with a hazard ratio of 2.003, a 95 percent confidence interval of 1.23 to 3.24, and a p-value of 0.005. A delay of more than three months carried a hazard ratio of 1.86, with a 95 percent confidence interval of 1.20 to 2.88, and the same p-value of 0.005. In practical terms, children whose diagnoses took longer than one or three months had roughly twice the hazard of failing treatment compared with those diagnosed sooner.

Relapse told a similar story. Diagnostic delays of more than one month were associated with relapse with a hazard ratio of 2.690, a 95 percent confidence interval of 1.19 to 6.34, and a p-value of 0.043, while delays of more than three months yielded a hazard ratio of 2.083, a 95 percent confidence interval of 1.08 to 3.96, and a p-value of 0.048. Notably, the Kaplan-Meier curves constructed for relapse on their own did not detect statistically significant associations; the significant relapse findings emerged only from the multivariable models adjusted for age and sex.

One result demands careful interpretation: the longest delays, beyond six months, were not significantly associated with non-response. The authors attribute this apparent discrepancy to a well-known epidemiological phenomenon called the waiting-time paradox, combined with a length-bias artifact. Children with highly aggressive, high-grade tumors tend to develop rapid, severe symptoms that prompt urgent medical evaluation, producing shorter diagnostic intervals but inherently worse biological outcomes. Children with slower-growing, low-grade tumors, by contrast, can survive long enough to accumulate extended pre-diagnostic intervals without showing immediate treatment failure. Tumor biology, in other words, confounds the relationship between delay and outcome at the extreme end of the distribution, and the authors caution that biological aggressiveness remains a key confounder when analyzing very long diagnostic delays.

The study’s limitations are substantial and the authors are candid about them. Determining the exact date of symptom onset retrospectively from electronic records is subjective and vulnerable to recall bias, since the researchers relied largely on clinical notes and caregiver reports that may be incomplete. This could have led to underestimation or misclassification of the true interval, potentially shifting some patients across the one-, three-, and six-month cutoffs and affecting the observed associations. The response assessment also carries caveats: although responses were evaluated six weeks after radiotherapy or surgery and after every three to four chemotherapy cycles depending on tumor type, the classification used for this study was based on the assessment at the end of the entire treatment course, and some patients may have been misclassified. The authors also note that RANO criteria remain bidimensional measurements and have not been validated against survival, and that the relationship between radiographic response and overall survival differs across pediatric CNS tumor types. The study was not designed or powered to detect differences in overall survival, follow-up is still ongoing, and the response rates reported should be read as an early sign of anti-tumor activity rather than a predictor of survival.

Context makes the findings consequential. Central nervous system tumors are the second most common childhood cancer after leukemia, and while five-year survival in some high-income countries ranges from 80 to 90 percent, it varies between 5 and 60 percent in low- and middle-income countries, with Mexican data indicating a long-term survival rate of about 50 percent. Unlike adult cancer, there are no evidence-based population screening programs or lifestyle risk-reduction strategies for childhood cancer, so timely diagnosis is one of the few levers available. Brain tumors in children are frequently misdiagnosed at first as migraines, gastroenteritis, or psychological and behavioral problems, and known contributors to delay include age at diagnosis, parental education level, cancer type, tumor location, and symptom presentation. A 2023 meta-analysis by Cotache and colleagues identified geographical, social, financial, health system, and policy barriers that restrict access to cancer care in low- and middle-income countries, though it did not assess the impact of delays on patients, and prior studies of delay and survival in pediatric solid tumors have produced inconsistent and contradictory results. The authors argue that emphasizing early recognition of these tumors at the primary care level, together with targeted training programs for primary care physicians and timely referral to hospitals capable of comprehensive treatment, could shorten the path to diagnosis and, if the associations they observed reflect a causal relationship, improve the odds that treatment works the first time.

Subject of Research: The association between diagnostic delay and treatment non-response and relapse in Mexican pediatric patients with central nervous system tumors

Article Title: Diagnostic delay as a prognostic factor for non-response to treatment and relapse in Mexican pediatric patients with central nervous system tumors

Article References: Pérez-De Marcos, J.-C., Castillejos-López, M., Aquino-Gálvez, A., Velasco-Hidalgo, L., Cárdenas-Cardos, R., Niembro-Zúñiga, A. M., De Uña-Flores, A., Zapata-Tarrés, M., & Torres-Espíndola, L.-M. (2026). Diagnostic delay as a prognostic factor for non-response to treatment and relapse in Mexican pediatric patients with central nervous system tumors. Frontiers in Pediatrics, 14, Article 1900510. https://doi.org/10.3389/fped.2026.1900510

Image Credits: AI Generated

DOI: 10.3389/fped.2026.1900510

Keywords: central nervous system tumors, pediatric oncology, diagnostic delay, treatment non-response, relapse, RANO criteria, Kaplan-Meier analysis, Cox regression, Mexico, glioma, waiting-time paradox, early diagnosis