BACKGROUND AND INTRODUCTION
Children and adolescents with cancer are surviving at increasing rates because of multimodal chemotherapy.1 One of the drugs frequently used in the treatment of pediatric cancers is vincristine, a vinca alkaloid. Children and adolescents treated with vincristine for cancer often develop chemotherapy-induced peripheral neuropathy (CIPN) during treatment.2,3 Although vincristine is an effective chemotherapeutic agent, it can also cause microtubule destabilization in noncancerous cells. In neurons, this leads to deficits in axonal transport and length-dependent neuropathy.4 Sensory, motor, and autonomic neuropathy as a result of vincristine exposure in children and adolescents has been reported.2,5–8 Motor deficits are seen frequently in both children during treatment2,9 and among survivors.5,7 Although it seems that acute neuropathy in many children and adolescents will lessen over time, some will be left with residual neuropathy that may have long-term functional implications.2,6,7,10 The effect of neurologic damage because of vincristine-containing therapy on spatiotemporal characteristics of gait has yet to be reported. Specific knowledge of the type of gait impairments seen in pediatric patients exposed to vincristine therapy may lead to better informed rehabilitation management in this population and could improve functional outcomes in survivors.
Thus, the primary purpose of this study was to describe spatiotemporal gait parameters in a pediatric oncology population with vincristine-associated peripheral neuropathy and to compare these findings with those of age- and sex-matched peers without cancer. For gait variables that demonstrated the greatest impairment in the CIPN population as compared with peers, we explored factors that may be important for this change in gait. Because anecdotally patients, families, and physical therapists have described gait abnormalities that increase with endurance tasks, we also investigated the effect of an endurance task (the 6-minute walk test [6MWT]) on the spatiotemporal aspects of gait in both groups. We hypothesized that gait deviations seen in subjects with CIPN would be exacerbated by the endurance task. Lastly, because decreased performance on 6MWTs has been described in patients with childhood cancer and survivors,6,11 we explored gait- and neuropathy-related factors that relate to performance on the 6MWT.
Children and adolescents 5 years of age or older being treated with vincristine for acute lymphoblastic leukemia (ALL), Hodgkin or non-Hodgkin lymphoma, or other solid non-central nervous system (CNS) tumors were recruited from the clinic population at 1 pediatric oncology center. Only children who scored 5 or greater on the pediatric-modified Total Neuropathy Score (ped-mTNS), indicating neuropathy,2 were included in this study of gait variables. Children and adolescents being treated for either CNS tumors, bony or muscle tumors of the lower extremity, or tumors impinging on the spinal cord were excluded. Children and adolescents with known osteonecrosis of the lower extremity at the time of gait evaluation were also excluded. Children and adolescents with ALL were tested at the end of their delayed intensification stage, which is approximately 6 months into treatment and occurs before the beginning of maintenance treatment. Subjects with Hodgkin or non-Hodgkin lymphoma or other solid tumors were tested near the end of their chemotherapeutic treatment; generally 3 to 6 months after chemotherapy began. Age- and sex-matched controls were recruited from siblings of patients attending the cancer and blood disorders clinic and children of staff members. All data were collected during a single visit. This study was approved by the Institutional Review Board at the participating institution. Participants younger than 18 years were required to give assent before study participation, and written consent was also obtained from their parent/guardian. Participants 18 years and over gave written consent before participation.
Height and weight were measured on all participants, and body mass index percentile was calculated using the CDC growth charts (www.cdc.gov/healthyweight/assessing/bmi/childrens_bmi/about_childrens_bmi.html). Leg length was measured in centimeters from the greater trochanter to the floor with the participant standing with weight equally distributed between the legs. Medical records were abstracted for cancer type and treatment variables including chemotherapy medications and their cumulative dosage, as well as radiation treatment and surgical procedures.
All participants (patients and controls) were tested for neuropathy using the pediatric-modified Total Neuropathy Score (ped-mTNS). The ped-mTNS is a combined subjective and clinical measure of peripheral nerve function with reported reliability and validity.2 It includes clinical examination of light touch, pin sensibility, vibration sensation, distal strength by manual muscle testing, and deep tendon reflexes (methodology described in Gilchrist and Tanner2). Scores of 5 or greater are more than 1 standard deviation from control population scores, and thus are used as markers of neuropathy.2 For patients with neuropathy, only children and adolescents undergoing chemotherapy for pediatric cancer, who had received the neurotoxic chemotherapeutic agent vincristine, and who had a ped-mTNS score of 5 or greater indicating CIPN, were included in this trial.
Trials of gait data were collected on a 14-ft portable GaitRite electronic walkway (CIR Systems, Sparta, New Jersey), with the participants walking barefoot at a comfortable, self-selected pace. This system consists of a roll-up carpet containing electronic pressure sensors and has demonstrated validity.12 The sensors detect footfalls, allowing for the calculation of temporal and spatial gait characteristics. The software program provides data on velocity (cm/s), cadence (steps/min), step length (cm), base of support (cm), and time in double support (percent of the cycle), as well as percent of time each portion of the foot (heel only, whole foot, forefoot only) is in contact with the pressure mat (Figure 1). Each participant completed 2 trials, and partial footfalls were excluded before analysis. Data from the 2 trials were averaged for analysis. After completing the 2 initial gait trials, all participants completed a 6MWT according to standardized published procedures13 and then immediately completed 2 more gait trials at a self-selected pace to assess for the effect of potential neuromuscular fatigue on gait.
Additional Measures of Impairments and Function
To assess the effect of specific impairments of body structures and function on gait, we additionally measured passive and active ankle dorsiflexion range of motion (ROM) as well as general strength and balance. Ankle ROM was measured with patients in the prone position, with the foot extending past the end of the table. Balance and strength subtests of the Bruininks-Oseretsky Test of Motor Proficiency Version 2 (BOT-2) were administered according to the standard procedure.14
Average step length (cm), cadence (steps/min), and velocity (cm/s) were all normalized to leg length (cm) according to Stansfield et al15 and in accordance to normal values for younger children published by Dusing and Thorpe.16 Data from 3 steps per trial over 2 trials were averaged for each individual, yielding the average of 6 steps per leg for each individual in each condition. If more than 3 steps were available for analysis from a given trial, the first 3 steps were used for analysis. Analyses were performed using SPSS version 20 (IBM, Armonk, New York). Parametric statistics were used after confirming normal distribution of data for all variables within each analysis group. Independent sample t tests were used to identify differences between groups. Associations between variables of interest were assessed using Pearson correlations. Multiple regression was used to assess the ability of the gait parameters and other physical impairments to predict results on the 6MWT and step length. Preliminary analysis was conducted to determine whether the models violated the assumptions of normality, linearity, multicollinearity, and homoscedasticity. The significance level was set at α < 0.05.
Of the 70 potential patients undergoing treatment for cancer that consented to participate in this study, 52 were included in this analysis. One was excluded due to technical issues with gait data collection, 4 had normal peripheral nerve function according to ped-mTNS testing, and 13 were excluded due to difficulty obtaining age- and sex-matched controls within the timeframe of the study. Eight of the 13 subjects excluded from analysis were aged 15 years or more, indicating the specific difficulty recruiting control subjects in this age range. The final population tested had an average age of 11.3 ± 4.5 years, ranging from 5 to 22 years, and 43.2% were males (Table 1).
Of the 52 patients undergoing cancer treatment, 23 (44%) were being treated for ALL, 18 (35%) were being treated for either Hodgkin or non-Hodgkin lymphoma, and 11 (21%) were being treated for another non-CNS solid tumor. Of the solid tumor group, 7 individuals were being treated for Wilms' tumor, and 4 individuals were being treated for rhabdomyosarcoma or Ewing sarcoma, 2 with each type. All patients received the neurotoxic agent vincristine, with a mean cumulative dose (± standard deviation) of 17.5 ± 8.8 mg/m2 and a range of 4.0 to 40.5 mg/m2. Many of the patients received other medications that may have additive or additional neurotoxic effects, including methotrexate and vinblastine. Thirty (58%) individuals received intrathecal methotrexate, with a mean cumulative dose of 132.3 mg and a range of 60.0 to 210.0 mg. Four individuals received vinblastine, with a mean cumulative dose of 18.8 mg/m2 and a range of 15.0 to 24.0 mg/m2. No individuals received platinum-based chemotherapy or bortezomib. The average time in treatment was 6.1 ± 2.4 months.
As shown in Table 1, no significant differences were found between groups in height or leg length. The patients undergoing treatment had significantly greater body mass index percentile scores. The mean neuropathy score for the patient group was significantly higher with a mean score of 10.6 ± 4.0 (range, 5-23) as compared with 1.2 ± 1.3 (range, 0-6) for controls (P < .001). Among patients, 92% (48 of the 52) of the subjects had grade 4 or worse ankle dorsiflexion strength, as compared with 8% of controls (4 of the 52). Abnormally high light touch thresholds at the great toe were found in 37% (19 of the 52) of patients and 6% of the control group (3 of the 52). Abnormally high vibration perception thresholds at the great toe were found in 46% of patients (24 of the 52) as compared with 2% (1 of the 52) of the control group. Deep tendon reflexes were reduced or absent at the ankle in 100% of patients and only 2% (1 of the 52) of the controls. Active and passive ankle dorsiflexion ROM was significantly lower in the patients undergoing cancer treatment (Table 1), with active ROM being less than passive ROM in both groups. The patients with neuropathy also had significantly lower scores on the balance and strength subscales of the BOT-2 (P < .001; Table 1).
When walking at a self-selected comfortable pace, the patients had significantly slower normalized velocity (4.0 ± 0.7) compared with controls (4.5 ± 0.8, P <.01). This difference in walking velocity seemed to be related primarily to a decreased step length (right 0.78 ± 0.10 cases vs 0.84 ± 0.09 controls, P < .01) rather than a decrease in cadence (312.9 ± 36.8 cases vs 317.9 ± 35.7, P = .48). The patients walked with a wider base of support than their age-matched peers (Table 2), yet the percent of the gait cycle spent in double support was not significantly different between groups (P = .71 right leg, P = .84 left leg). The percent time each portion of the foot (heel only, whole foot, forefoot only) was examined at self-selected walking velocity and no significant difference was seen between groups (Table 2).
Table 3, columns 1 and 2 show the correlations between specific impairments and step length and gait velocity during the pre-6MWT trials. The factors of step length and cadence demonstrated the highest correlations with normalized velocity. Impairments correlated with the outcome of interest at 0.30 or greater were entered into the multiple linear regression models to evaluate the independent contributions to these outcomes. Preliminary analysis for normalized velocity indicated a major deviation for normal distribution of the residual values, even after data transformation. Because step length is highly correlated with walking velocity and differs between subjects with CIPN and controls, we then investigated the relative contributions of different factors to step length. The 2 impairments with unique contributions to step length were the balance subscale score (β = 0.31, P = .001) and ankle passive dorsiflexion ROM (β = 0.25, P = .009) (F(3,101) = 12.88, R2 = .20, P < .001). Neuropathy scores and strength subscale scores on the BOT-2 did not add unique contributions to the variance. Whereas the total ped-mTNS score did have a correlation greater than 0.3 with step length, none of the individual items (sensory, strength, or reflex scores) met this benchmark.
After the gait assessment, all subjects completed a 6MWT to assess not only their functional ambulation capacity, but also to create the potential for neuromuscular fatigue. Patients, on average, did not walk as far in 6 minutes as compared with controls (498.6 ± 88.2 m vs 604.3 ± 86.2, P < .01) and had significantly lower age- and sex-specific Z-scores (−2.4 ± 1.7 vs −0.6 ± 1.3, P < .01).17 After the 6MWT, the self-selected velocity during gait evaluation increased for both groups, though the patients remained significantly slower than the controls (4.5 ± 0.8 vs 5.2 ± 0.8, P < .01). After the 6MWT, step length increased in both groups, though the patient group continued to have shorter step length than the controls (Table 2). Cadence, which was not different between groups before the 6MWT (Table 2), was significantly slower among patients than among controls after the 6MWT (329.6 ± 35.6 vs 353.2 ± 37.4, P < .01). Base of support remained wider in the CIPN group, though the control group slightly increased their width from pre-6MWT values (Table 2). Time in double support did not change appreciably for either group. Analysis of changes in foot contact with the mat surface during gait demonstrated that percent time the heel was in contact with the mat decreased for the control group after the 6MWT, whereas the time the forefoot was in contact with the mat increased for the CIPN group after the 6MWT (Figure 1, Table 2). This yielded a net increase in the time the entire foot was in contact with the ground for the subjects with CIPN as compared with the control group after the 6MWT (Table 2).
Multiple regression was also used to assess the ability of the pre-6MWT gait parameters and other physical impairments to predict results on the 6MWT. Performance on the strength subscale test on the BOT-2 demonstrated the strongest contribution (β = 0.43, P < .001) to 6MWT Z-score variability, whereas scores on the ped-mTNS (β = −0.29, P = .001) and self-selected normalized gait velocity before the 6MWT (β = 0.18, P = .024) also contributed to the variability (F(4,100) = 37.74, R2 = 0.53, P < .001). The variables of active or passive ankle ROM and balance did not improve the fit of the model or add unique contributions to explaining the variance.
In this sample of children and adolescents with peripheral neuropathy because of vincristine exposure, self-selected walking velocity was slower than age- and sex-matched peers. This slower velocity seems to be due to a decreased step length and wider base of support and not due to decreased cadence as no significant difference was found between patients and controls for this variable. The slowed walking velocity found in this sample is similar to findings in other groups of subjects with peripheral neuropathy from other causes. Wuehr et al18 found in adult patients with peripheral neuropathy from multiple causes, but not chemotherapy drug exposure, that walking velocity was decreased. In their adult population, both cadence and stride length were decreased,18 yet in our sample cadence was not significantly different from controls. In a study of adults with diabetic peripheral neuropathy, Martinelli et al19 also found decreased gait speed and decreased stride length as compared with age-matched healthy controls. In their study, stride length and gait speed were associated with both decreased ankle ROM and isometric dorsiflexion and plantar flexion muscle strength. In a small pilot study of patients with pediatric cancer and survivors, Wright et al20 documented impairments in muscle activation and timing in 11 children and adolescents either on-treatment or off-treatment for ALL who had evidence of peripheral neuropathy. In their subjects, abnormalities were documented in the timing and activation of the tibialis anterior and the gastrocnemius muscles, with one of the resulting effects being decreased power generation at push-off. The data of Wright et al20 are consistent with our finding of decreased step length and impairments in distal muscle strength measured on the ped-mTNS, as decreased power at push-off would affect step length more than cadence.
When the factors underlying the decrease in step length were investigated using regression analysis, we found that balance scale scores and tightness in ankle dorsiflexion best explained our variance. Tightness in ankle dorsiflexion, clearly prevalent in the sample of patients with cancer tested here and reported in other studies,5,21 is often hypothesized by practicing physical therapists to lead to early heel rise. This did not seem to be the case in a large proportion of our sample, although early heel rise still may be seen in a few individuals. Possibly to accommodate the reduced ankle ROM most of the children in our study reduced step length rather than allowing their heel to rise earlier. In addition to the association between ankle tightness and step length, we found an association between impaired balance and reduced step length. Although neuropathy was not an independent factor in determining step length variability, likely neuropathy and balance are closely related constructs as has been previously demonstrated in children with non-CNS cancer,2 and thus would not be revealed as 2 independent factors. Indeed for neuropathy to affect gait variables, the neuropathy may need to be sufficiently severe to affect balance. Interestingly, an overall measure of neuropathy (the ped-mTNS score) was more highly correlated with gait variables than either strength or sensation scores alone, indicating that both sensation and strength are likely influential on gait parameters.
In our experience, families of children treated for cancer report that their children tripped more frequently after beginning to tire when ambulating in the community. Thus, we evaluated changes to gait parameters pre- and post-6MWT, hypothesizing that neuromuscular fatigue may become evident after an endurance task. Interestingly, after the 6MWT, both those with neuropathy and controls had increased walking velocity as compared with pre-6MWT trials, indicating an effect of recent fast walking on self-selected walking velocity. In both groups, step length and cadence increased during the post-6MWT trials, although patients with neuropathy had shorter average step lengths than controls both pre- or post-6MWT. When comparing the post- to the pre-6MWT gait evaluation, controls spent less of each step with the heel in contact with the ground, and approximately the same percent of the step with the forefoot in contact with the ground. The patients with CIPN maintained the same percent of the step with the heel in contact with the ground, but increased the percent of the step with the forefoot in contact with ground. This change in forefoot contact may reflect decreased eccentric control and neuromuscular fatigue of weakened ankle dorsiflexors in children with vincristine-related neuropathy. The timing of footfall characteristics has been rarely investigated in peripheral neuropathy, but our results are in accord with the findings of Abboud et al,22 who demonstrated a faster forefoot contact or “forefoot slap” related to late firing of the tibialis anterior in a sample of adults with diabetes.
Functional walking capacity as measured by distance in the 6MWT was also shown to be limited in this sample with peripheral neuropathy and is consistent with data from survivors of childhood leukemia5 as well as in children with a recent diagnosis of cancer.11 In our sample, general strength impairments as described by the BOT-2 subscale, as well as neuropathy scores and self-selected “normal” walking speed before the 6MWT, were associated with decreased performance on the 6MWT. Although children and adolescents with peripheral neuropathy could increase their walking speed when asked on the 6MWT, and increased self-selected walking velocity on the post-6MWT, the subjects with CIPN walked slower than their age- and sex-matched peers on every gait assessment.
Although many of the gait characteristic reported in our sample of children and adolescents with CIPN from vincristine use are consistent with changes in gait parameters seen in other populations with neuropathy, reasons other than vincristine neuropathy may contribute to our results. First, many of the patients received additional medications such as intrathecal methotrexate that could potentially increase in neuropathy signs and symptoms in a portion of this sample. We have recently found that patients with ALL, who are most frequently treated with intrathecal methotrexate, typically have lower neuropathy scores than patients with other types of cancer.23 Decreased performance on the 6MWT could be due not only to peripheral neuropathy, but also related to disease or treatment-induced muscle wasting and cardiac and pulmonary effects of treatment. Although none of the children with cancer had known cardiotoxicity at the time of the testing, subclinical cardiac changes most likely existed in a portion of the sample24 and deconditioning is likely, even early in treatment, in children with these cancers.11 Although testing a sample of children with pediatric cancer who did not develop neuropathy may have mitigated this effect, the high prevalence of neuropathy in children treated at our clinic made such a comparison not feasible. It also should be noted that the gait evaluation reported here is based on footfall data taken at 1 time point during treatment, and not based on a full kinematic analysis. Thus, other gait alterations may have been present in this sample. We must also emphasize that the gait changes noted in this study occur at a single time point after a period of intense medical treatment, and it is unknown whether these changes persist or require remediation. Thus, further research is warranted.
Children and adolescents undergoing cancer treatment who develop vincristine-related CIPN demonstrate altered spatiotemporal characteristics of gait that are related to functional changes in walking capacity. Ankle dorsiflexion ROM and balance impairments are important factors for therapists to consider when treating step length deficits in this population, whereas strength is also an important consideration when intervening for decreased functional walking capacity.
The authors thank Ann Logelin for data abstraction, Jessica Ovans, PT, and Katherine Wacker, PT, for assistance with data collection, and Dr David Chapman and Dr Kirsten Ness for their consultation on the article.
1. Ward E, DeSantis C, Robbins A, Kohler B, Jemal A. Childhood and adolescent
cancer statistics, 2014. CA Cancer J Clin. 2014;64(2):83–103. doi:10.3322/caac.21219.
2. Gilchrist LS, Tanner L. The pediatric-modified total neuropathy score: a reliable and valid measure of chemotherapy-induced peripheral neuropathy in children with non-CNS cancers. Support Care Cancer. 2013;21(3):847–856.
3. Smith EML, Li L, Hutchinson RJ, et al. Measuring vincristine-induced peripheral neuropathy in children with acute lymphoblastic leukemia. Cancer Nurs. 2013;36(5):E49–E60.
4. Park SB, Krishnan AV, Lin CS, et al. Mechanisms underlying chemotherapy-induced neurotoxicity and the potential for neuroprotective strategies. Curr Med Chem. 2008;15(29):3081–3094.
5. Ness KK, Hudson MM, Pui CH, et al. Neuromuscular impairments in adult survivors of childhood acute lymphoblastic leukemia: associations with physical performance and chemotherapy doses. Cancer. 2012;118(3):828–838. doi:10.1002/cncr.26337.
6. Ness KK, Jones KE, Smith WA, et al. Chemotherapy-related neuropathic symptoms and functional impairment in adult survivors of extracranial solid tumors of childhood: results from the St. Jude Lifetime cohort study. Arch Phys Med Rehabil. 2013;94:1451–1457.
7. Jain P, Gulati S, Seth R, et al. Vincristine-induced neuropathy in childhood ALL (acute lymphoblastic leukemia) survivors: prevalence and electrophysiological characteristics. J Child
8. Ramchandren S, Leonard M, Mody RJ, et al. Peripheral neuropathy in survivors of childhood acute lymphoblastic leukemia. J Periph Nerv Syst. 2009;14(3):184–189.
9. Gilchrist LS, Marais L, Tanner L. Comparison of two chemotherapy-induced peripheral neuropathy measurement approaches in children. Support Care Cancer. 2014;22(2):359–366.
10. Khan RB, Hudson MM, Ledet DS, et al. Neurologic morbidity and quality of life in survivors of childhood acute lymphoblastic leukemia: a prospective cross-sectional study. J Cancer Surviv. 2014;8(4):688–696.
11. Ness KK, Kaste SC, Zhu L, et al. Skeletal, neuromuscular and fitness impairments among children with newly diagnosed acute lymphoblastic leukemia. Leuk Lymphoma. 2014;20:1–8.
12. Bliney B, Morris M, Webster K. Concurrent related validity of GaitRite walkway system for quantification of spatial and temporal parameters of gait. Gait Posture. 2003;17:68–74.
13. American Thoracic Society. ATS statement: guidelines for the six-minute walk test. Am J Resp Crit Care Med. 2002;166:111–117.
14. Bruininks RH, Bruininks BD. BOT 2: Bruininks-Osteresky Test of Motor Proficiency, 2nd edn, Manual. Minneapolis, MN: Pearson Assessments; 2005.
15. Stansfield BW, Hillman SJ, Hazlewood ME, et al. Normalisation of gait data in children. Gait Posture. 2003;17:81–87.
16. Dusing SC, Thorpe DE. A normative sample of temporal and spatial gait parameters in children using the GAITRite electronic walkway. Gait Posture. 2007;25(1):135–139.
17. Geiger R, Strasak A, Tremel B, et al. Six-minute walk test in children and adolescents. J Pediatr. 2007;150:395–399.
18. Wuehr M, Schniepp R, Schlick C, et al. Sensory loss and walking speed related factors for gait alterations in patients with peripheral neuropathy. Gait Posture. 2014;39:852–858.
19. Martinelli AR, Mantovani AM, Nozabieli AJL, et al. Muscle strength
and ankle mobility for the gait parameters in diabetic neuropathies. Foot. 2013;23:17–21.
20. Wright MJ, Twose D, Gorter JW. The functional impact of peripheral neuropathy in children and youth treated for acute lymphoblastic leukemia: a feasibility study. Meeting of the Gait and Clinical Movement Analysis Society; 2013.
21. Marchese VG, Chiarello LA, Lange BJ. Effects of physical therapy intervention for children with acute lymphoblastic leukemia. Pediatr Blood Cancer. 2004;42(2):127–133.
22. Abboud RJ, Rowley DI, Newton RW. Lower limb muscle dysfunction may contribute to foot ulceration in diabetic patients. Clin Biomech. 2000;15(2):37–45.
23. Gilchrist L, Tanner L. Chemotherapy-induced peripheral neuropathy in non-CNS cancers: comparison between diagnostic groups. Congress of the International Society of Paediatric Oncology; 2014.
24. Lipshultz SE, Cochran TR, Franco VI, Miller TL. Treatment-related cardiotoxicity in survivors of childhood cancer. Nat Rev Clin Oncol. 2013;10(12):697–710.