Effects of proactive and rescue enteral tube feedings on weight change in children undergoing treatment for high-grade CNS tumors

Abstract

Background

Children with high-grade CNS cancers frequently experience malnutrition during treatment. We assessed the effects of proactive enteral tube (ET) placement/enteral tube feedings (ETF) on weight in infants/children with high-grade CNS tumors treated with aggressive chemotherapy.

Methods

We conducted a retrospective study of patients age 0 to 19 years treated for new high-grade CNS tumors between 2002 and 2017 at a tertiary pediatric hospital system. Patients underwent placement of proactive ET (≤ 31 days postdiagnosis; n = 45), rescue ET (> 31 days, due to weight loss; n = 9), or no ET (n = 18). Most received surgically placed ET (98%), with percutaneous endoscopic gastrojejunostomy or gastrojejunostomy tubes favored to allow jejunal feeding. The majority of patients with ET used ETF (91%). Using mixed-effects regression models, we examined differences in mean weights between ET/ETF groups across the first year of treatment. We also evaluated observed weight changes.

Results

All infants (n = 22, median age, 1.5 years) had proactive ET placed and 21 of 22 used proactive ETF. Infants showed an initial increase in mean percentage weight change that eventually leveled off, for an estimated increase of 10.4% over the year. For the pediatric cohort (n = 50, median, 8.1 years), those receiving proactive ETF experienced weight increases (+9.9%), those with rescue ETF experienced an initial decline and eventually rebounded for no net change (0.0%), and those with no ETF demonstrated an initial decline that persisted (–11.9%; P_interaction < .001). Analysis of observed weights revealed nearly identical patterns.

Conclusions

Proactive ETF was effective at maintaining weight and/or facilitating weight gain over the first year of treatment and was acceptable to patients/families.

CNS tumors are the second most common cancer in children age 0 to 14 years and the third most common cancer in adolescents age 15 to 19 years, accounting for 21% and 10% of cases, respectively.¹ Medulloblastoma and tumors formerly grouped as supratentorial primitive neuroectodermal tumors (sPNET)² together constitute the most common malignant pediatric CNS tumors and account for nearly 20% of all childhood CNS cancers.³ The current standard therapy for pediatric medulloblastoma/sPNET includes safe maximal surgical resection of the tumor followed by aggressive chemotherapy with or without radiation therapy. Infants (age < 3 years) with high-grade glioma, atypical teratoid rhabdoid tumor (ATRT), choroid plexus carcinoma, and ependymoma often receive aggressive treatment regimens very similar to those used for infant medulloblastoma/sPNET. Although cure rates for pediatric CNS cancers have improved over time,^4–6 in part because of the addition of intense chemotherapy and radiation therapy, the aggressive treatments are accompanied by substantial side effects. Malnutrition is a common complication of cancer treatment in pediatric patients and is attributable to multiple factors, including nausea, vomiting, constipation, altered taste, decreased appetite, food aversion, mucositis, and swallowing dysfunction. Malnutrition has been shown to adversely affect the child’s energy level, tolerance to treatment, risk for infection, neurodevelopment, linear growth, and other quality-of-life measures.⁷

Malnutrition is an overlooked and understudied complication of aggressive treatment for pediatric CNS cancers. To date, only a few publications have investigated the causes, effects, and potential interventions for malnutrition seen in children treated for medulloblastoma/sPNET.^1,8,9 These studies demonstrated significant weight loss during treatment for children with medulloblastoma/sPNET and superior nutritional outcomes in patients who received supplemental enteral feeding.^1,8 In addition, there is a growing body of literature validating the effectiveness of enteral feedings in improving nutrition in pediatric oncology patients who experience weight loss during treatment^8,10–12 and several studies indicating proactive enteral feedings, initiated before significant weight loss, are most effective in maintaining nutrition and weight in children undergoing cancer treatment.^1,13–15

To prevent the untoward effects of malnutrition, Children’s Minnesota began recommending proactive enteral tube (ET) placement in 2002 for infants younger than 3 years undergoing treatment for newly diagnosed high-grade CNS cancers and in 2011 for children 3 years and older undergoing treatment for newly diagnosed medulloblastoma/sPNET with the goal of maintaining weight and good nutrition throughout treatment. Here we report on the effectiveness of proactive and rescue (ie, in response to weight loss) ET placement and enteral tube feeding (ETF) initiation on maintaining and improving the weight of infants (age < 3 years) with high-grade CNS tumors and children (age ≥ 3 years) with medulloblastoma/sPNET who received aggressive treatment at our center.

Methods

We conducted a retrospective cohort study within a tertiary children’s hospital system to assess the effect of ET and ETF on weight change in infants, children, and adolescents over the course of treatment for newly diagnosed high-grade CNS malignancies. The study protocol was approved by the institutional review board before data collection commenced. Parents/legal guardians provided written consent for their child’s data to be included in research studies at the time of their initial institutional encounter and annually thereafter. In addition, investigators received a waiver of consent for the current chart review study.

Our study objective was to examine a set of uniform malignant diagnoses for patients who were well nourished at diagnosis and who received aggressive chemotherapy (plus or minus radiotherapy) regimens that are associated with a high likelihood of dietary malnutrition (due to decreased appetite, nausea, etc). Given this objective, patients were eligible if they were treated at age 3 to 19 years for a newly diagnosed medulloblastoma or sPNET or at age younger than 3 years for a newly diagnosed high-grade CNS tumor (medulloblastoma, sPNET, high-grade glioma, ATRT, choroid plexus carcinoma, or ependymoma), and completed treatment with aggressive combination chemotherapy according to selected protocols (SJYC07 [NCT00602667], SJMB12 [NCT01878617], CCG-99701 [NCT00003203], CCG-99703 [NCT00003141], COG-A9961 [NCT00002875], COG-ACNS0331 [NCT00085735], COG-ACNS0332 [NCT00392327]) or equally aggressive treatment regimen at Children’s Minnesota during the period 2002 to 2017 (Fig. 1). Patients may or may not have received radiation in the course of their treatment. Patients were excluded if their parent/guardian did not consent to research, if the treatment received was not eligible, or if they died before the end of treatment. In addition, we excluded patients treated on infant protocols who did not undergo proactive ET placement, but received rescue ETF (n = 1) or no ETF (n = 2) because the sample sizes in those groups prohibited statistical analysis and interpretation.

Fig. 1.

Flow Diagram of Patient Eligibility and Inclusion. ETF, enteral tube feedings; sPNET, supratentorial primitive neuroectodermal tumors.

If a patient’s family agreed to proactive ET placement, a nasogastric (NG), nasojejunal (NJ), gastrostomy (G), gastrojejunostomy (GJ), percutaneous endoscopic gastrostomy (PEG), or percutaneous endoscopic gastrojejunostomy (PEG-J) tube was placed at the time of diagnosis. Our standard ET recommendations included surgically/endoscopically placing an ET because of the expected more than 6-month need for ETF throughout treatment and the high incidence of nasal enteral tubes dislodging and requiring replacement; placing an ET that allowed for administration of jejunal feeds (ie, GJ or PEG-J tube) because of the high incidence of vomiting in this patient population; and placing the ET at the time that the patient underwent anesthesia for placement of a central venous line to avoid a second sedation. PEG-J tubes were the predominant surgically placed GJ tube according to the preference of our gastroenterology team.

At diagnosis, a registered dietitian met with each family. Families were educated on the importance of good nutrition in promoting healthy weight and development throughout treatment, as well as common nutrition intervention needs. Anthropometrics were measured initially and throughout the duration of treatment. For patients who had an ET placed, the dietitian provided written and verbal instructions to the family on their child’s caloric, protein, and fluid needs and how to use supplemental enteral feeds (bolus vs continuous drip vs both) to make up for any insufficient dietary intake. Adjustments were made over the course of treatment based on oral intake and weight. Patients who did not have an ET placed were also monitored by the dietitian throughout treatment. Attempts were made to promote appropriate weight maintenance or gain, such as the use of nutrition supplements and appetite stimulants. If these attempts were unsuccessful, ETF was recommended.

ETF was recommended at diagnosis (for the infant group in 2002-2017 and pediatric group with proactive ET after 2011) or if weight loss was more than 5% to 10% of baseline body weight (pediatric group before 2011 and any pediatric patient after 2011 who did not have a proactive ET). Exposure to ETF was therefore classified in 2 ways: with respect to the timing of ET placement and the timing of ETF initiation. All patients treated on infant protocols and included in the present analysis received proactive ET placement (n = 22) and all but one received proactive ETF (n = 21). The remaining infant received rescue ETF in response to weight loss. Patients treated on pediatric protocols were categorized as receiving proactive ET placement (ET placed within 31 days of diagnosis; n = 23), rescue ET placement (ET placed > 31 days after diagnosis in response to weight loss; n = 9), or no ET placement (n = 18). Because some pediatric patients had an ET placed proactively or as a rescue effort but did not initiate feedings (ETF) until later or not at all, and some pediatric patients did not have enough weight loss to warrant ETF, pediatric patients were further categorized as having received proactive ETF (ETF initiated within 31 days of diagnosis; n = 20), rescue ETF in response to weight loss (ETF initiated > 31 days after diagnosis; n = 8), or no ETF despite weight loss and recommendation to start ETF by the dietitian (n = 11). Notably, the latter classification (ETF initiation) involved recategorization of 4 patients according to actual ETF use and did not include the 11 patients for whom ETF was not recommended by the dietitian because of lack of significant weight loss.

Data were abstracted from electronic medical records into a standardized REDCap electronic abstraction form¹⁶ by trained abstractors and practitioners (C.R.B., R.M.L., K.M.I., and A.E.B.). Abstracted fields included CNS malignancy diagnosis, age at diagnosis, sex, treatment regimen, weight (in kilogram) and date at diagnosis, presurgery, postsurgery, preradiation, postradiation, before each cycle of chemotherapy, and at the completion of all treatment (ie, 30-days postchemotherapy follow-up), date of (initial and subsequent) ET placement, type of ET, and complications from ET placement. Data on complications related to ET/ETF were collected for the duration of the patient’s ET course, including ET site erythema, ET site pain, oral aversion, and other complications. Sex-specific Z scores for weight at the time of diagnosis were generated using the CDC Growth Charts^17,18 via the PediTools Online Calculator (https://peditools.org/). As a final quality assurance step, the abstracted data underwent central review by the neuro-oncologist (A.E.B.) and registered dietitian (R.M.L.).

Statistical Analysis

To examine differences across ET placement/utilization groups, we used chi-square tests or Fisher exact tests when cell sizes were less than 10 to compare proportions, t tests or F tests to compare means, and a nonparametric Wilcoxon rank sum/Kruskal-Wallis test to compare medians and distributions.

We evaluated differences in mean weight and mean percentage weight change (calculated as 100% *[(weight at time point – baseline weight)/baseline weight]) separately by treatment protocol type (infant vs pediatric protocols) and ET subgroup (proactive, rescue, or no ET placement; proactive, rescue, or no ETF initiation). We examined the differences in the means at different time points using 2 different methods.

First, we constructed mixed-effects regression growth models to examine percentage weight change across the first year of treatment by subgroup. This method is useful because patients did not follow the exact same course of treatment: Radiation was administered at different points in the course of treatment and different numbers of chemotherapy cycles were administered overall. Importantly, the method is robust to missing data.

For the infant protocol group, the final model included a quadratic function of day of treatment (modeled as day + day²), which was included both as a fixed and a random effect. For the pediatric protocols, by which we sought to examine differences between the proactive, rescue, and no ET placement/initiation groups across the treatment time period, we modeled the cubic function of day of treatment (modeled as day + day² + day³) and the interaction terms between the cubic function and treatment group as fixed effects; the intercept term and cubic function of day of treatment were modeled as random effects. The 6 degree-of-freedom test for interaction from this model is therefore a global test of whether the growth curves for the different groups are different from one another. From these results, we plotted both the predicted growth curves from the mixed-effects regression models and the observed data points for those treated according to infant (Fig. 2) and pediatric protocols (Fig. 3) to allow for visualization of the data across the treatment period.

Fig. 2.

Predicted Mean Percentage Weight Change (Line) via Mixed-Effects Growth Modeling and Observed Percentage Weight Change (Data Points) During First Year After Diagnosis in Patients Treated on an Infant Protocol Who Received Proactive Enteral Tube (ET) Placement (n = 22).

Fig. 3.

Predicted Mean Percentage Weight Change (Line) via Mixed-Effects Growth Modeling and Observed Percentage Weight Change (Data Points) During First Year After Diagnosis by Timing of A, Enteral Tube (ET) Placement (n = 50) and B, Enteral Tube Feeding (ETF) Initiation (n = 39) in Patients Treated on a Pediatric Protocol For data points, circles indicate proactive. ET/ETF, triangles indicate rescue ET/ETF, and squares indicate no ET/ETF.

Second, we looked at mean weight differences based on treatment milestones (after – before radiation, last – first chemotherapy, fourth round of chemotherapy – first round, and 30-day postchemotherapy follow-up – diagnosis). This method is useful when considering the effects of specific treatment modalities (eg, radiation) and the total effects from diagnosis to last follow-up. We applied statistical tests to determine whether observed differences were significantly different from zero (1-sample t tests) or from other ET subgroups (F tests and 2-sample t tests, for those treated on pediatric protocols only).

Data analysis was performed via Stata Statistical Software: Release 14 (StataCorp LP, 2015). In all statistical tests, 2-sided P values less than .05 signified statistically significant differences.

Results

A total of 72 patients were ultimately included in our retrospective cohort study (Fig. 1), including 22 treated according to infant protocols and 50 treated according to pediatric protocols during the period 2002 to 2017. Of these, 54 patients (75%) had ETs placed and 49 (68%) received enteral nutrition (Table 1). Proactive ETs were placed in 45 patients (62.5%) and rescue ETs were placed in 9 (12.5%). Of patients who had ETs placed, 49 (91%) used ETF per recommendations by the dietitian and 5 (9%) did not. Of the remaining 18 patients who did not have an ET placed, 7 (39%) had weight loss during treatment but the families refused ETF and 11 (61%) maintained nutrition without the need of ETF.

Table 1.

Selected Characteristics for Patients Treated on Infant and Pediatric Protocols, Respectively, by Enteral Tube (ET) Placement Group

	Infant Protocol	Pediatric Protocol
	Proactive ET a	Proactive ET	Rescue ET	No ET	P
	(n = 22)	(n = 23)	(n = 9)	(n = 18)
	No. (%)	No. (%)	No. (%)	No. (%)
Age, y
Median (IQR)	1.5 (1, 2)	7 (4, 10)	8 (7, 11)	9 (4, 10)	.54
≥ 3	3 (13.6)	23 (100.0)	9 (100.0)	18 (100.0)
Male sex	10 (45.5)	13 (56.5)	6 (66.7)	13 (72.2)	.59
CNS tumor
Medulloblastoma	11 (50.0)	22 (95.7)	8 (88.9)	16 (88.9)	.52
sPNET	3 (13.6)	1 (4.4)	1 (11.1)	2 (11.1)
Ependyoma	4 (18.2)	0	0	0
ATRT	3 (13.6)	0	0	0
Glioma	1 (4.6)	0	0	0
Z score for weight at diagnosis
Median (IQR)	0.66 (–0.53, 1.14)	–0.22 (–0.57, 0.47)	0.67 (–0.48, 0.85)	0.63 (–0.19, 1.17)	.09
Z score for weight at last follow-up
Median (IQR)	0.22 (–0.17, 0.66)	–0.37 (–1.10, 0.12)	0.07 (–0.54, 0.31)	–0.21 (–0.72, 0.72)	.53
Treatment protocol
SJMB12	0	5 (21.7)	2 (22.2)	1 (5.6)	.003
COG-9961	0	2 (8.7)	0	6 (33.3)
COG-ACNS0331	0	7 (30.4)	5 (55.6)	6 (33.3)
COG-ACNS0332	0	8 (34.8)	0	0
CCG-99701	0	1 (4.4)	2 (22.2)	2 (11.1)
SJYC07	15 (68.2)	0	0	0
CCG-99703	3 (13.6)	0	0	0
Other	4 (18.2)	0	0	3 (16.7)
Radiation
Focal only	11 (50.0)	0	0	0	–
Focal plus craniospinal	2 (9.1)	23 (100.0)	9 (100.0)	18 (100.0)
Radiation type
Photon	5 (22.7)	12 (52.2)	7 (77.8)	15 (83.3)	.10
Proton	8 (36.4)	11 (47.8)	2 (22.2)	3 (16.7)
Initial ET placedb
NG tube	3 (13.6)	4 (17.4)	1 (11.1)	0	.97
NJ tube	1 (4.6)	2 (8.7)	1 (11.1)	0
PEG/G tube	2 (9.1)	2 (8.7)	1 (11.1)	0
PEG-J/GJ tube	16 (72.7)	15 (65.2)	6 (66.7)	0
Compliance
ETF recommended	22 (100.0)	23 (100.0)	9 (100.0)	7 (38.9)	< .001
ETF used as instructed	21 (95.5)	20 (87.0)	8 (88.9)	NA	1.00
Length of follow-up (months)
Median (IQR)	10.0 (8.7, 13.5)	10.7 (9.5, 15.0)	14.9 (10.4, 15.2)	15.0 (10.9, 16.2)	.04

Abbreviations: ATRT, atypical teratoid/rhabdoid tumor; CCG, Children’s Cancer Group legacy protocol; COG, Children’s Oncology Group protocol; ET, enteral tube; ETF, enteral tube feeding; G, gastrostomy; GJ, gastrojejunostomy; IQR, interquartile range; NA, not available; NG, nasogastric; NJ, nasojejeunal; PEG, percutaneous endoscopic gastrostomy; PEG-J, percutaneous endoscopic gastrojejunostomy; SJMB, St. Jude medulloblastoma protocol; SJYC, St. Jude young child protocol; sPNET, supratentorial primitive neuroectodermal tumor.

^aExcludes 3 patients who did not have proactive ET placement: rescue ET, n = 1; no ET, n = 2.

^bSome patients started with an NG or NJ tube until surgical placement of a GJ tube could be scheduled.

Infant Protocols

The 22 patients treated on infant protocols SJYC07, CCG-99703, and similar regimens had a median age of 1.5 years (interquartile range, IQR: 1.0, 2.0 years) at diagnosis; 46% of them were male, 50% were treated for medulloblastoma, and the remainder were treated for sPNET (14%), ependymoma (18%), ATRT (14%), or glioma (5%; Table 1). Of note, 3 patients with medulloblastoma treated on SJYC07, considered an infant protocol, were age 3 years or older but younger than 5 years. Notably, only 59% of patients received therapeutic radiation at any time during the course of their treatment. The 22 patients had a median weight Z score of 0.66 (IQR: –0.53, 1.14) at diagnosis, suggesting that they were not undernourished at the time of diagnosis/ET placement.

Visual inspection of the best-fitting mixed-effects regression curve for the first year following diagnosis shows a monotonic increase in average percentage weight change through day 269 and a leveling off thereafter, for a predicted net increase of 10.4% (Fig. 2). A minority of patients continued to exhibit negative changes (as illustrated by the data points in Fig. 2). These results are consistent with the weight changes we calculated across treatment milestones, including a statistically significant 13% increase between day 1 of the first and last chemotherapy sessions (P < .001), and a 15% increase overall between the date of diagnosis and the 30-day postchemotherapy follow-up visit (P < .001; Table 2). The median weight Z score for the infant group at the end of treatment was 0.22 (IQR: –0.17, 0.66), which was similar to their distribution at diagnosis and indicated that the infants had gained an appropriate amount of weight in relation to their linear growth.

Table 2.

Observed Mean Weights and Percentage Weight Change Across Treatment Milestones by Type of Protocol (Infant vs Pediatric) and Timing of Enteral Tube Placement

	Infant Protocol	Pediatric Protocol
	Proactive ETa (n = 22)	Proactive ET (n = 23)	Rescue ET (n = 9)	No ET (n = 18)	P
	Mean (SD)	Mean (SD)	Mean (SD)	Mean (SD)
Weight, kg
Diagnosis	12.7 (3.5)	28.6 (15.5)	33.6 (13.5)	34.6 (17.1)	.45
30-day postchemotherapy follow-up	14.2 (3.1)	28.9 (12.2)	33.4 (12.0)	34.0 (14.1)	.41
Percentage weight change
Before versus after radiation	0.09 (6.6)	2.8 (9.6)c	–2.1 (7.0)c,d	–6.0 (7.0)b,d	.007
First versus last chemotherapy	13.0 (11.7)b	5.3 (12.2)b	11.2 (7.7)b	7.0 (10.2)b	.39
First versus fourth round chemotherapy	5.5 (10.6)b	3.5 (10.5)	7.1 (7.3)b	3.6 (10.1)	.63
Diagnosis versus follow-up	14.7 (12.5)b	6.3 (18.9)	1.4 (7.5)	2.4 (13.6)	.64

Abbreviation: ET, enteral tube.

^aOnly 13 out of 22 patients in the infant protocol group had radiation therapy.

^b P was less than .05 in 1-sample t test, indicating the percentage weight change was statistically distinguishable from 0.

^c,dMean percentage differences with the same lowercase letter are not statistically significantly distinguishable from one another, whereas values with different letters are significantly different with P less than or equal to .05.

Because 21 of the 22 patients treated on an infant protocol underwent ETF as instructed, results for ET placement (Table 2) and proactive initiation of ETF (Table 3) were very similar.

Table 3.

Observed Mean Weights and Percentage Weight Change Across Treatment Milestones by Type of Protocol (Infant vs Pediatric) and Timing of Enteral Tube Feeding Initiation

	Infant Protocol	Pediatric Protocol
	Proactive ETF a (n = 21)	Proactive ETF (n = 20)	Rescue ETF (n = 8)	No ETF (n = 11)	P
	Mean (SD)	Mean (SD)	Mean (SD)	Mean (SD)
Weight, kg
Diagnosis	12.3 (3.1)	25.8 (11.3)c	34.2 (14.3)c,d	41.9 (20.7)d	.02
30-day postchemotherapy follow-up	14.0 (2.9)	28.0 (11.3)	33.8 (12.8)	35.9 (15.4)	.24
Percentage weight change
Before versus after radiation	0.4 (6.9)	3.6 (10.0)c	–2.1 (7.5)c,d	–6.5 (7.3)b,d	.01
First versus last chemotherapy	14.2 (10.7)b	7.2 (11.4)b	10.4 (7.8)b	1.0 (11.0)	.15
First versus fourth round chemotherapy	6.9 (8.6)b	4.5 (10.8)	5.7(6.5)b	0.6 (11.7)	.51
Diagnosis versus follow-up	15.6 (12.0)b	10.6 (15.8)b,c	0.8 (7.7)c,d	–10.9 (11.4)b,d	< .001

Abbreviation: ETF, enteral tube feeding.

^aOnly 12 out of 21 patients in the infant protocol group had radiation therapy.

^b P was less than .05 in 1-sample t test, indicating the percentage weight change was statistically distinguishable from 0.

^c,dMean percentage differences with the same lowercase letter are not statistically significantly distinguishable from one another, whereas values with different letters are significantly different with P less than or equal to .05.

Minor ET complications were commonly observed during at least one visit during treatment among patients treated on an infant protocol who underwent proactive ET placement. Of the 22 patients, 10 (45.5%) had at least one episode of cellulitis or erythema at the ET site and 9 (40.9%) experienced pain at the ET site, whereas no patients had a severe infection at the ET site. One infant suffered an intussusception at the distal jejunostomy tube site, which resolved following removal of the jejunostomy tube and did not recur with continued feeding through the gastrostomy tube. Six (27%) developed short term reversible oral aversion.

Pediatric Protocols

With respect to the 50 patients treated on pediatric protocols (SJMB12, CCG-99701, COG-A9961, COG-ACNS0331, and COG-ACNS0332) and similar aggressive regimens, 64% were male, 92% were treated for medulloblastoma, and the rest were treated for sPNET (Table 1). All patients received radiation (craniospinal and focal boost) during the course of treatment, with 68% receiving photon radiation and 32% proton radiation. ETF was used as instructed in 28 (72%) of the 39 pediatric patients in whom ETF was recommended (proactive ET [n = 23], rescue ET [n = 9], and no ET [n = 7]). On stratification by timing of ET placement, the only notable difference observed across the 3 groups was the median Z score for weight at the time of diagnosis. As might be expected, patients who received an ET proactively had a somewhat lower Z score distribution at diagnosis (n = 23; median: –0.22, IQR: –0.57, 0.47) than those who received an ET as a rescue measure (n = 9; median: 0.67, IQR: –0.48, 0.85) and those who did not have an ET placed (n = 18; median: 0.63, IQR: –0.19, 1.17; P = .09).

Considering first the timing of ET placement, visual inspection of the smoothed percentage weight change curves (Fig. 3A) indicates that the trajectories for weight change over the first year of treatment for the 3 groups were different. The proactive group showed an initial modest decrease in weight change with a nadir at 22 days (–0.5%), followed by a period of growth that reached a maximum at 318 days and culminated in a net gain of 5.5% at day 365. The rescue group showed an initial decline that extended to 96 days (–6.9%) followed by an increase that peaked at 311 days and finished the year with a net increase of 0.5%. In contrast, the no ET group experienced a notable decline through day 150 (–5.1%) followed by a rebound so there was a net weight increase at the end of the year (2.1%). The test for interaction revealed that the global differences in the 3 curves reached borderline statistical significance (P_interaction = .07).

In examining weight changes across different treatment milestones, statistically significant differences in percentage weight change before vs after radiation were observed across the 3 ET placement groups (P = .007; Table 2; Supplementary Fig. 1A), in which the proactive ET group gained an average of 2.8%, whereas the no ET group lost 6.0%; the –2.1% change in the rescue ET group was indistinguishable from the differences in the other 2 groups. No other significant differences were found across groups and notably, the observed weight changes from diagnosis through the last follow-up visit were not distinguishable from 0 in any of the 3 groups.

Next, we looked at the effects of ETF utilization by examining only those patients who had ETF recommended, either because the ET was placed proactively or because of weight loss. In this analysis, a similar pattern was observed for the proactive and rescue ETF groups as was described previously for ET placement; however, the curve for the no ETF group (ie, ETF recommended because of weight loss but ETF refused) was substantively different (Fig. 3B). Specifically, the proactive ETF group exhibited a monotonic increase in weight over the first year of treatment (+9.9%), whereas the rescue ETF group experienced an S-shaped curve that included an initial decline through day 94 (–7.2%) followed by an increase through day 314 (+1.6%) and a final decline resulting in no net change at 1 year (0.0%), and the no ETF group underwent an initial decline through day 141 (–11.7%) that remained throughout the year, for a net decrease of 11.9% at day 365. The test for interaction corroborated these results because a significant difference between the growth curves was detected (P_interaction = .0009).

The milestone data for ETF utilization also showed somewhat different results from the ET placement results. Notably, larger differences were observed from diagnosis to the 30-days postchemotherapy follow-up visit in the ETF utilization analysis (P < .001; Table 3; Supplementary Fig. 1B). The proactive ETF group experienced an increase in weight over the entire period (+10.6%), whereas the rescue ETF group showed little change (+0.8%) and the no ETF group demonstrated an average decline (–10.9%). These changes resulted in median weight Z scores that were indistinguishable for the proactive (–0.27, IQR: –0.76, 0.31), rescue (–0.15, IQR: –0.93, 0.29) and no ETF (–1.03, IQR: –1.66, 0.57) groups at the 30-day post-chemo follow-up visit (P = .60). Notably, these Z scores indicate that our patients were similar at the end of treatment to children of the same age in the general population.

Of patients treated on pediatric protocols who received proactive ET placement (n = 23), 15 (65.2%) experienced at least one episode of erythema at the ET site and 11 (47.8%) reported at least one episode of pain at the ET site. Patients who had rescue ET placement (n = 9) had complication rates that were statistically indistinguishable from those in patients receiving proactive ET placement. Four patients (44.4%) had one or more episodes of erythema at the insertion site and 2 (22.2%) reported pain at the site. No patients had a severe infection at the ET site or radiation-induced colitis, enteritis, or esophagitis. In considering only the 28 patients who received ETF as recommended, 6 (21%) experienced short-term, reversible oral aversion.

Discussion

In this retrospective cohort study conducted within a single pediatric tertiary care system, infants with newly diagnosed high-grade CNS cancers and children with newly diagnosed medulloblastoma or sPNET who received proactive ETF experienced consistent increases in weight over the first year of treatment. In contrast, those who received rescue ETF secondary to weight loss underwent an initial decline in weight and eventually rebounded after the initiation of ETF so there was no net change in weight detectable over the course of treatment, whereas those who had ETF recommended because of weight loss, but did not use ETF, demonstrated an initial weight loss that was not recovered by the end of follow-up.

Poor nutrition and weight loss are commonly seen during treatment for childhood cancer, with significant weight loss demonstrated in as many as 60% of all pediatric cancer patients combined¹⁹ and in 46% to 88% of pediatric patients with high-grade CNS cancers undergoing aggressive treatment.^8,9,13 Our results are consistent with others’ observations of children undergoing treatment for high-grade CNS cancers, during which patients generally have normal weight at diagnosis, but then lose weight during the radiation¹ and chemotherapy phases of treatment.^1,8 Our results are also consistent with those from other studies that have shown beneficial effects of ETF initiated at the time of weight loss during pediatric cancer treatment, resulting in stabilization or improvement in weight.^{1,8,10,12–15,19,20} Unlike other reports, we did not use parenteral nutrition, but instead opted to use ETF because ETF has been demonstrated to be superior to parenteral nutrition at improving weight in pediatric patients undergoing treatment for medulloblastoma and sPNET.⁸ There is a growing body of literature that supports proactive ETF as the superior way to maintain and improve weight in children undergoing treatment for high-grade CNS tumors^8,13 and for other malignancies.^13–15 Our study confirms the benefit of proactive ETF in high-grade CNS cancers because patients who received proactive ETF were the only group who experienced consistent increases in weight over the course of treatment.

Our study has a few unique features compared to other reported studies assessing ETF utilization in pediatric cancer patients. We included 45 patients who underwent placement of a proactive ET and 41 patients who used proactive ETF, which is the largest number of patients reported for each. Because we looked both at the number of patients who underwent ET placement (proactive or rescue) and the number of patients who used ETF (proactive or rescue) we were able to assess the compliance of ETF initiation once an ET was placed, with 49 of the 54 patients (91%) using ETF to provide adequate daily caloric needs as recommended by the dietitian. This compliance rate suggests that families are amenable to using ETF once an ET is in place and therefore supports proactive placement of an ET at the time of diagnosis if the patient is at a high risk for malnutrition. Of note, given the fact that we recommended utilization of ETF at the first sign of decreased oral intake in our proactive ET patients, it was not feasible to determine how nutrition and weight might have fared in these children if they had used oral nutrition alone, but the observations that the proactive ETF patients had no significant weight loss and that their weight gain was superior during cancer treatment compared to the rescue ETF and no ETF groups also support proactive ETF utilization in this patient population. Despite our preference to place a proactive ET, 41% of our pediatric patients who did not undergo placement of a proactive ET did not have significant weight loss during treatment and therefore had no need for an ET. Therefore, another viable option would be to delay ET placement, with diligent surveillance and initiation of ETF at the first sign of poor oral nutrition or minimal weight loss. Of interest, no parent expressed regret or provided other negative feedback in response to having an ET placed. Instead, all parents found the ET helpful for ease of administering medications, fluids, and nutrition, including the 5 of 54 patients (9%) who had an ET placed but did not use it for enteral feedings.

Another unique feature of this study was the high rate of surgically placed ETs over nasal ET and the high rate of placement of GJ or PEG-J tubes over G tubes or PEG tubes to allow for jejunal feedings. At our center we prefer using surgically placed ET over nasal ET because of the lower risk of dislodgement and need for replacement and because we believe they are more cosmetically acceptable. We also prefer placing a GJ or PEG-J because it allows the option to use either the G or the J portion of the tube, with the G tube used for bolus and continuous drip feeds and administration of medications and the J tube used for continuous drip feeds if nausea and vomiting are an issue. To this point, only one patient used only an NG tube throughout treatment, whereas the other 53 patients with ETs eventually received surgically placed GJ or PEG-J tubes.

Previous reports confirm our observation that surgically placed ETs have minimal side effects when used in pediatric cancer patients during treatment.^{12,13,19,21–23} Of interest, Schmitt et al reported that the rate of complications seen with use of a PEG tube in pediatric cancer patients was no higher than that seen in pediatric patients using a PEG tube because of neurological impairment.¹⁴ We saw no major complications with the surgically placed ETs, and the minor complications seen were temporary and consisted mainly of ET site inflammation and pain, typically occurring during periods of neutropenia. Oral aversion was seen at similar frequencies in the infant (27%) and pediatric (21%) ETF groups, which was a concern for some parents and sometimes required the assistance of a feeding clinic, but all cases ultimately resolved.

Several reports have warned that surgical placement of an ET in a patient with a ventriculoperitoneal (VP) shunt might increase the risk for VP shunt infection,²¹ therefore Gassas and colleagues recommended delaying surgical ET placement until 6 weeks following placement of a VP shunt.²⁴ In our study, 19 patients with surgically placed ET (35%) also required placement of a VP (17), subdural (1), or lumbar shunt (1), with 12 (63%) shunts placed before the ET (range, 3-105 days, median, 20.5 days) and 7 (27%) placed after the ET (range, 4-94 days, median, 11 days). There were no cases of shunt infection.

A strength of our study is that we were able to analyze our observational study data in 2 complementary ways: We applied an intention-to-treat analysis (ET placement) and also looked at the effectiveness of our intervention in a real-world setting (ETF utilization). We saw important, and not unexpected, differences between the results of the 2 analyses because some patients who received a proactive ET or rescue ET did not use it as prescribed.

There were also a few limitations to our study worthy of mention. Because it was a retrospective observational study involving home administration of the enteral feedings, we were not able to capture exactly how much nutrition was delivered via tube feeding to assess effectiveness. Instead, we evaluated effectiveness based on the recommendations of the dietitian and whether the dietitian stated that the family was compliant with ETF recommendations. Because all patients treated on infant protocols had ETs placed prospectively, there were no contemporaneous rescue or no ET/ETF comparison data available; we assume infants with rescue or no ET/ETF would have had results similar to those observed in older children. Other limitations included the relatively modest numbers of patients in each ET and ETF subgroup and the heterogeneity in treatment modalities and timelines, which required us to average across many variables in our analysis, potentially confounding or masking our true results. Whether ETF affects survival outcomes is an interesting question that cannot be readily addressed in the present study, but could be addressed in a larger prospective study.

Overall, our results suggest that proactive ET placement with proactive EFT is effective at maintaining weight over the course of radiation and chemotherapy for pediatric high-grade CNS cancers and is acceptable to most patients/families. Although a randomized controlled trial would be needed, strictly speaking, to confirm the results of our study, expenditure of the resources required for such a trial seem unnecessary given the acceptability, type of minor complications encountered, reasonable cost,^10,11 and risk-to-benefit ratio of the ETF intervention. We will continue with our practice of proactive ETF in infants and children with medulloblastoma, sPNET, and other high-grade CNS tumors that require aggressive therapeutic interventions.

Funding This work was supported by Children’s Minnesota.

Conflict of interest statement: None declared.