Separable systems for recovery of finger strength and control after stroke

Abstract

Impaired hand function after stroke is a major cause of long-term disability. We developed a novel paradigm that quantifies two critical aspects of hand function, strength, and independent control of fingers (individuation), and also removes any obligatory dependence between them. Hand recovery was tracked in 54 patients with hemiparesis over the first year after stroke. Most recovery of strength and individuation occurred within the first 3 mo. A novel time-invariant recovery function was identified: recovery of strength and individuation were tightly correlated up to a strength level of ~60% of estimated premorbid strength; beyond this threshold, strength improvement was not accompanied by further improvement in individuation. Any additional improvement in individuation was attributable instead to a second process that superimposed on the recovery function. We conclude that two separate systems are responsible for poststroke hand recovery: one contributes almost all of strength and some individuation; the other contributes additional individuation.NEW & NOTEWORTHY We tracked recovery of the hand over a 1-yr period after stroke in a large cohort of patients, using a novel paradigm that enabled independent measurement of finger strength and control. Most recovery of strength and control occurs in the first 3 mo after stroke. We found that two separable systems are responsible for motor recovery of hand: one contributes strength and some dexterity, whereas a second contributes additional dexterity.

Keywords: finger individuation; motor recovery; plasticity; strength; stroke.

Copyright © 2017 the American Physiological Society.

Figures

Fig. 1.

Strength and individuation task. Fifty-four patients and 14 healthy controls were tested 5 times over a 1-yr period. A: ergonomic hand device. Force transducers beneath each key measured the force exerted by each finger in real time. The participant’s fingers are securely placed on the keys using Velcro straps. B: computer screen showing the instructional stimulus, which indicates both which finger to press and how much force to produce (height of the green bar). In the MVF task, maximal force was required. MVF trials were performed twice on each finger. In the individuation task, 1 of 4 target force levels had to be reached. Target force levels were 20%, 40%, 60%, and 80% of MVF for each finger. Individuation trials were performed 4 times per force level per digit. C and D: example trials from 2 healthy control participants during the individuation task. Four trials are shown, one at 20% and one at 80% of MVF for the 2 participants. In this case the fourth finger (inset, red) was the instructed finger. Note the higher level of enslaving of the uninstructed fingers for higher instructed finger force level. E: mean deviation from baseline in the uninstructed fingers plotted against the force generated by the instructed finger for C and D. The Individuation Index is the −log (slope) of the regression line between instructed finger force and uninstructed mean deviation, measured as root mean square (RMS) force from baseline force produced by uninstructed fingers. F and G: example trials from 1 patient during the individuation task, two trials from each time point over the 1-yr period at 40% (F) and 80% MVF (G). The instructed finger for each trial was the same as those shown in control data (C and D). H: force-control trade-off function at each time point for the example patient, showing higher level of enslaving early after stroke and recovery of individuation ability over time.

Fig. 2.

Temporal profiles of recovery for strength and individuation. A and B: group recovery curves for the Strength and Individuation Indexes for patients and controls. Asterisks indicate significant week-to-week changes for the paretic hand [Bonferroni corrected P values for each segment of Strength Index: P(W1–W4) = 0.0045, P(W4–W12) = 0.0082, P(W12–W24) = 0.068, and P(W24–W52) = 0.87; Individuation Index: P(W1–W4) = 0.81, P(W4–W12) = 0.0024, P(W12–W24) = 1.92, and P(W24–W52) = 2.91]. C: rate of change (i.e., change per week) in z-normalized Strength and Individuation Indexes during the first 2 intervals (W1 to W4 and W4 to W12). The 2 intervals show a significant interaction between strength and individuation, indicating faster initial improvement of strength. D: week-to-week correlations between adjacent time points for the Strength and Individuation Indexes. Dashed lines are the noise ceilings based on the within-session split-half reliabilities (sample size N = 53 patients, 14 controls, with a total of 251 completed sessions).

Fig. 3.

Temporal recovery profiles for clinical measures: Fugl-Meyer for the arm (A; FM-Arm) and hand (B; FM-Hand) and the ARAT (C). All measures showed significant change over time for the paretic hand. FM-Arm: χ2 = 37.73, P = 8.13 × 10−10; FM-Hand: χ2 = 29.03, P = 7.14 × 10−8; ARAT: χ2 = 36.33, P = 1.67 × 10−9 (sample size N = 53 patients, 14 controls, with a total of 238 and 235 completed sessions for FM and ARAT, respectively).

Fig. 4.

Time-invariant impairment function relating strength and control. A: scatter plots for Individuation vs. Strength Indexes at each time point. Each black circle is one patient’s data; blue circles and ellipse indicate the mean and SE for controls at the time point. Two patients’ data are highlighted: one with good recovery (yellow circle) and one with poor recovery (red circle). The total number of patients at each time points is 39, 39, 40, 39, and 34 for W1–W52, respectively. B: scatter plot with data from all time points superimposed with the best-fitting two-segment piecewise linear function with one inflection point at Strength Index = 0.607. C: residuals from each week, with the mean impairment function (red line in B) subtracted. The tendency of the residuals to stay above or below the typical Strength-Individuation relationship indicates that there are stable factors that determine individuation recovery over and above strength recovery. D: correlations of residuals from C across adjacent time points increased over time [Bonferroni-corrected P(W1–W4) = 2.12, P(W4–W12) = 0.00064, P(W12–W24) = 0.0024, and P(W24–W52) = 3.39 × 10−6]. Dashed line is the noise ceiling based on the within-session split-half reliabilities (sample size N = 53 patients, 14 controls, with a total of 251 completed sessions).

Fig. 5.

Schematic diagram of the hypothesis for two recovery systems. The first system (basic strength recovery) underlies strength recovery and a restricted amount of individuation recovery. This system therefore defines the lower bound (dashed line) of the space occupied by recovering patients (gray shading). A second system (additional individuation recovery; vertical arrows) adds further individuation abilities on top of basic strength recovery.

Fig. 6.

Lesion distribution and correlation with behavior. A: averaged lesion distribution mapped to JHU-MNI space (see materials and methods), with lesion flipped to one hemisphere. Color bar indicates patient count. B: correlation of behavior measures (Strength and Individuation Indexes) at each time point with the percentage of damaged cortical gray matter within the M1 hand area ROI and the corticospinal tract (CST). Black asterisks indicate significant correlations (tested against 0), and red asterisks indicate a significant difference (P < 0.005) between the correlation for strength and individuation for each week. C: mean of week-by-week correlations between the two behavior measures and percent lesion volume measures for the cortical gray matter hand area and CST ROIs (sample size N = 53, with 251 completed sessions).