Change in knee cartilage components in stroke patients with genu recurvatum analysed by zero TE MR imaging

Genu recurvatum in stroke patients with hemiplegia causes readily cumulative damage and degenerative changes in the knee cartilage. It is important to detect early cartilage lesions for appropriate treatment and rehabilitation. The purpose of this cross-sectional study was to provide a theoretical basis for the early rehabilitation of hemiplegia patients. We used a zero TE double-echo imaging sequence to analyse the water content in knee joint cartilage at 12 different sites of 39 stroke patients with genu recurvatum and 9 healthy volunteers using a metric similar to the porosity index. When comparing the hemiplegic limb vs. the nonhemiplegic limb in patients, the ratios of the deep/shallow free water content of the femur cartilages at the anterior horn (1.16 vs. 1.06) and posterior horn (1.13 vs. 1.25) of the lateral meniscus were significantly different. Genu recurvatum in stroke patients with hemiplegia can cause changes in the moisture content of knee cartilage, and the changes in knee cartilage are more obvious as the genu recurvatum increases. The "healthy limb" can no longer be considered truly healthy and should be considered simultaneously with the affected limb in the development of a rehabilitation treatment plan.

Approximately 70% of stroke patients have hemiplegia, of whom 40-60% have knee hyperextension 1 . Long-term hyperextension of the knee changes the mechanical trend of the lower limbs, and the knee joint load-bearing reaction worsens such that the stability of the standing phase decreases and the body leans forwards. These changes induce an uneven distribution of the internal stress in the knee joint, cause knee pain and cartilage injury when walking in this way for a long time, and lead to cumulative damage and degenerative changes in knee cartilage 2 . Other tissues around the knee joint, such as the meniscus, also undergo pathological changes 3 . Therefore, there is no doubt that hemiplegic patients with knee reflexes will develop early knee injury. At present, there is no effective drug to treat cartilage injury and repair cartilage defects 4 . If the lesion is detected by imaging before irreversible damage occurs to the articular cartilage and appropriate treatment is adopted, normal knee function can be maintained, and disability can be avoided. Therefore, it is important to detect early lesions of the cartilage.
Cartilage is mainly composed of chondrocytes and extracellular matrix. The extracellular matrix is mainly composed of water, collagen fibres and proteoglycans, and water accounts for 80% of the wet weight of cartilage 5 . In osteoarthropathy, the initial histological and biochemical changes in cartilage are collagen fibre breakdown, decreased proteoglycan content, and increased water dispersion 6 . Therefore, early cartilage lesions manifest as an increased moisture percentage, which is associated with the loss or damage of matrix components 7 . The detection of moisture changes in cartilage would be helpful for the early detection of cartilage lesions.
Although the gold standard for imaging diagnosis in osteoarthropathy is X-ray examination with the Kellgren-Lawrence grading system, it cannot be used to directly assess cartilage. Recently, MRI has become a powerful tool for the examination of early changes in tissue components in vivo and has been used to evaluate osteoarthropathy 8,9 . Traditional MRI uses T2WI and T2 mapping to image cartilage 10,11 , but these MRI sequences can only image long TE tissues and the superficial and intermediate layers of articular cartilage, while the deep and calcified layers cannot be imaged 12 13 . Ultrashort TE sequences overcome this problem by imaging both long and short TE tissues 14 . Recently, many scholars have used these sequences to study articular cartilage, especially UTE T2* mapping and T1 rho mapping, which can quantitatively analyse early changes in endochondral injury [15][16][17][18] . However, these techniques are difficult to apply in clinical practice because of their complexity, long scanning time and the influence of the magic angle effect. Some scholars used the porosity index of the double-echo UTE sequence to evaluate the porosity of the bone cortex, calculated as the ratio of the image intensities of the second echo (indicating signals from pore water) and the first echo (indicating signals from all water), and found it to be highly consistent with the results measured by micro-CT 19,20 . This method is relatively simple and has potential for clinical use. The contrast and signal-to-noise ratio of short TE tissue imaging with a zero TE sequence are basically the same as those of the ultrashort TE sequence 21 . The difference between them is that the readout gradient of the ultrashort TE sequence starts at the beginning of data acquisition, while that of the zero TE sequence starts before the radiofrequency pulse.
No literature has reported changes in early knee cartilage in stroke patients with hemiplegia and knee retraction on MRI. In this study, a zero TE double-echo imaging sequence was used to analyse the water content in the knee joint cartilage of stroke patients with genu recurvatum using a metric similar to the porosity index, and the study focused on three questions: (1) is there any difference between the knee joint cartilage of the hemiplegia side and the nonhemiplegia side in stroke patients? (2) Are the changes in knee cartilage related to the degree of genu recurvatum? (3) Is there any difference between the knee cartilage of stroke patients and normal people? This study aims to provide a theoretical basis for the early rehabilitation of hemiplegia patients.

Patients. A total of 151 stroke patients with hemiplegia received rehabilitation treatment in the Neurological
Rehabilitation Center of the Beijing Rehabilitation Hospital from January 2019 to October 2020. A total of 39 patients with complete data met the study criteria, including 30 males and 9 females, ranging in age from 32 to 60 years, with an average age of 49 years, and a body mass index (BMI) ranging from 21 to 31 kg/m 2 . Among the 39 patients, three had no available gait data, 9 did not have genu recurvatum (< 5°), 15 had mild genu recurvatum (5°-10°) and 12 had severe genu recurvatum (> 10°). There were 9 healthy volunteers, 5 males and 4 females, ranging in age from 35 to 50 years old, with an average age of 42.5 years and a BMI between 20 and 26 kg/m 2 . There were no significant differences in age, BMI or sex distribution between stroke patients and healthy volunteers (P > 0.05). The morphology and signal of the knee cartilage and meniscus were normal on conventional MRI. The paralyzed limbs, time of illness and walking duration of the patients were recorded (Table 1). Regardless of the handedness and the side of the paralyzed limb, all patients received loading practice and gait training for both limbs, since they were both affected.
Volunteer versus patient. Among the patients with right limb hemiplegia, the affected limbs were compared with the corresponding limb of healthy volunteers. The deep free-water content of the tibia cartilage at the anterior horn and posterior horn of the lateral meniscus and the femur cartilage at the anterior horn of the medial meniscus of the affected limb were decreased. The content of free water in the deep layer of the tibial cartilage at the posterior horn of the medial meniscus was decreased, and the content of binding water in the shallow layer was also decreased. However, the free water content in the shallow layer and the binding water in the deep layer of the femur cartilage at the posterior horn of the lateral meniscus of the affected limb was also decreased. The content of free water in the load-bearing cartilage of the medial femoral condyle of the affected limb was relatively increased. When the relatively healthy limb of the patients with right limb hemiplegia was compared with the corresponding limb of healthy volunteers, the deep binding water of the tibia cartilage at the posterior horn of the medial meniscus of the relatively healthy limb was reduced. At the femur cartilage at the posterior horn of the medial meniscus, the free water content in the shallow layer of cartilage was decreased, while the free water content in the deep layer increased and the combined water was also decreased. The free water content in the shallow layer of the femur cartilage at the posterior horn of the lateral meniscus was decreased. The deep free water content in the load-bearing cartilage of the medial femoral condyle of healthy limbs was relatively high.
When comparing the affected limb of the patients with left limb hemiplegia with the corresponding limb of healthy volunteers, we found that the deep binding water of the tibia cartilages at the anterior horn and posterior horn of the medial meniscus and the femur cartilage at the posterior horn of the medial meniscus of the affected limb were decreased. The free water content in the shallow layer of the femur cartilage at the posterior horn of the lateral meniscus, the tibia cartilage at the posterior horn of the medial meniscus and the load-bearing area of the lateral condyle of the tibia was decreased. When the relatively healthy limbs of the patients with left limb hemiplegia were compared to the corresponding limbs of healthy volunteers, the deep free water content of the femur cartilage at the anterior horn of the medial meniscus, the tibia cartilage at the anterior horn and posterior www.nature.com/scientificreports/ horn of the lateral meniscus, and the tibia cartilage at the posterior horn of the lateral meniscus of the relatively healthy limb were lower, and the free water content of the shallow layer of the tibia cartilage at the anterior horn of the lateral meniscus was also lower. The free water content in the shallow and deep layers of cartilage in the lateral condyle of the tibia of the relatively healthy limbs was decreased.
The changes in the above indicators were statistically significant at P < 0.05 ( Fig. 1).

Hemiplegic limb vs. nonhemiplegic limb.
There were significant differences in the ratio of the deep/ shallow free water content of the femur cartilage at the anterior horn and posterior horn of the lateral meniscus and the deep/shallow binding water content of the tibia cartilage at the anterior horn and posterior horn of the medial meniscus (P < 0.05) ( Table 2). The deep free-water content of the tibia cartilage at the anterior horn of the medial meniscus and that of the cartilage at the posterior horn were higher in the nonhemiplegic limbs, while the combined water content of cartilage was higher in the hemiplegic limbs.
Comparison between patients with and without genu recurvatum. In the affected limbs, the content of free water in the shallow layer of the tibia cartilage at the anterior horn of the lateral meniscus and the femur cartilages at the anterior horn and posterior horn of the lateral meniscus were higher in the patients with genu recurvatum than in those with no genu recurvatum. In patients with genu recurvatum, the deep binding water of the femur cartilage at the anterior and posterior horns of the lateral meniscus was higher than in the patients with no genu recurvatum, while the deep binding water of the tibia cartilage at the posterior horn of the medial meniscus was lower than in the patients with no genu recurvatum. In addition, the deep free water of the femur cartilage at the posterior horn of the medial meniscus was lower than that in patients with no genu recurvatum, while the load-bearing cartilage at the lateral condyle of the tibia was higher than that in patients with no genu recurvatum.
In the healthy limbs, the deep free water content of the tibia cartilage at the anterior horn of the lateral meniscus, the femur cartilage at the anterior horn of the medial meniscus and the load-bearing region of the lateral condyle of the tibia cartilage in patients with genu recurvatum were higher than those in patients with no genu recurvatum. In patients with genu recurvatum, the free water in the shallow layer of the femur cartilage at the posterior horn of the medial meniscus and the femur cartilage at the posterior horn of the lateral meniscus were higher than in patients with no genu recurvatum. The deep binding water of the femur cartilage at the posterior horn of the lateral meniscus was higher than that in patients with no genu recurvatum (Fig. 2).

Comparison between patients with mild and severe genu recurvatum.
In the affected limbs, the deep free water content of the femur cartilages at the anterior and posterior horns of the medial meniscus and the load-bearing area of the lateral condyle of the femur in the patients with severe genu recurvatum were greater than those in patients with mild genu recurvatum. The deep binding water of the femur cartilages at the anterior horn of the medial meniscus and the load-bearing areas of the medial femoral condyle and medial tibial condyle were lower than those of patients with mild genu recurvatum. The free water content in the shallow layer of the tibia cartilage at the anterior horn of the lateral meniscus was less than that in the patients with mild genu recurvatum.
In the healthy limbs, the free water content of the femur cartilage shallow layer at the posterior horn of the medial meniscus in the patients with severe genu recurvatum was higher than that in patients with mild genu recurvatum, and the deep free water content of the femur cartilages at the anterior horn of the lateral meniscus and the load-bearing area of the medial condyle of the femur were higher than those in patients with mild genu recurvatum.
The changes in these indices were statistically significant at P < 0.05 (Fig. 3).

Discussion
The preliminary results showed that the knee cartilage of stroke patients had significant changes compared with the normal group, and the changes were more significant with increased genu recurvatum degree; both the healthy limb and the affected limb showed cartilage changes. Articular cartilage is divided into four layers, starting from the articular surface, the superficial layer, the intermediate layer, the deep layer and the calcified layer, below which is the subchondral bone. Conventional magnetic resonance imaging cannot distinguish the cartilage calcification layer from subchondral bone 22,23 and cannot accurately measure cartilage changes. As subchondral bone is the same as the bone cortex, the T1 relaxation time is short, and the proton density is low 24 , but the cartilage calcification layer is different; although it has the same short T1 relaxation time as subchondral bone, the proton density is high 25 . Although the proton density is very high, the T1 relaxation time of the upper deep cartilage is very long. Therefore, the calcification layer on UTE appears as a high signal line above the subchondral bone 26 (Fig. 4). The high signal line is used to separate the cartilage layer from the subchondral layer, which avoids the influence of subchondral bone on the cartilage signal intensity and improves the accuracy of measurement.
Some MR techniques have been designed to evaluate the substance and its concentration [27][28][29] . For articular cartilage, there are three kinds of water molecules that can be investigated by the zero TE and UTE techniques. Eighty percent of the water is free water, with a long T2 value between 130 and 145 ms; 3% is bound to proteoglycan, with a T2 value between 8 and 12 ms; and 12% of the water is bound to collagen, with a T2 value between 3 and 18 ms. Multicomponent T2 analysis of articular cartilage 30 yields similar results, with 6% bound to collagen, 14% bound to proteoglycan, and 80% free. The long T2* components obtained by UTE two-component analysis correspond to free water, while short T2* components indicate proteoglycan-bound water and collagen-bound water 26  www.nature.com/scientificreports/ corresponds to free water. Therefore, the changes in free water and bound water, such as those in extracellular matrix water, proteoglycan and fibrinogen at different stages of osteoarthropathy, could be obtained by comparison between the two. The water content in the surface layer of normal cartilage was abundant and decreased to the deep layer of cartilage, while the degree of aggregation of proteoglycans was the opposite 31 . The results showed that the content of binding water in the deep layer was higher than that of the shallow layer in all parts of the cartilage, and the free water content of the shallow layer in most parts of the cartilage was higher than that of the deep layer. The femoral cartilage at the anterior horn and posterior horn of the lateral meniscus or medial meniscus were the only locations in which the free water content of the deep layer was higher than that of the shallow layer. This might be related to stress, but the specific reasons need to be further studied.
From the onset of osteoarthritis, the proteoglycan and collagen fibrous network continues to undergo degenerative changes. During the development of osteoarthritis, the three kinds of extracellular matrix change correspondingly. The water content increases in the early stage of the disease and decreases in the middle and late stages. Proteoglycan rapidly decreases from the onset of osteoarthritis to later stages of the disease. Highly ordered collagen fibres become disordered in the early stage and seriously broken in the late stage [32][33][34][35] .
The affected and healthy limbs of hemiplegia patients were compared with the corresponding limbs of healthy volunteers. Both the affected and healthy limbs showed statistically significant differences in different parts of the cartilage, which basically showed a decrease in shallow free water content with or without a decrease in binding water or an increase or decrease in deep free water content and a decrease in binding water. These changes in the early and middle stages were consistent with the degeneration of articular cartilage. The different locations might be related to the changes in the mechanics of the affected and healthy limbs and the different mechanics of the cartilage, which need to be further explained in conjunction with mechanical analyses. Many studies have found significant asymmetry between hemiplegic limbs and nonhemiplegic limbs in stroke patients. The hemiplegic limbs have a slow gait rate, and their dysfunction leads to short standing durations, prolonged swinging durations, and reduced ground responses [36][37][38] , leading to insufficient power and poor progress. To adapt to this compensatory dysfunction, patients need to rely on the nonhemiplegic limb to maintain balance and push forwards, resulting in biomechanical changes of the nonhemiplegic limb, presenting with an erratic gait 39 . Patients prefer using the nonhemiplegic side limb to bear weight, resulting in asymmetry and interruption of the hemiplegic gait 40 . Therefore, healthy limb degenerative changes occur during the compensatory process, which was also confirmed by our experimental results. There were significant differences in some parts of the cartilage in the healthy limbs between the patients with hemiplegia and the healthy volunteers, which indicated that the so-called "healthy limbs" of the patients had also changed to a certain extent and could not be called truly healthy limbs. Therefore, in the process of making rehabilitation treatment plans, the "healthy" limbs should be considered together with the affected limbs.
In stroke patients, the femur cartilages at the anterior and posterior horn of the lateral meniscus and the tibia cartilages at the anterior and posterior horn of the medial meniscus in the knee of the healthy limb were different from those in the affected limb, indicating that the mechanical mechanics and walking modes of the healthy limb in patients were different to compensate for the function of the affected limb, leading to different degrees of damage.
Compared with that in patients with no genu recurvatum, the cartilage in both affected and healthy limbs of patients with genu recurvatum was significantly altered, showing increased shallow and/or deep free water, but the deep free water in the femur cartilage at the posterior horn of the medial meniscus of the affected limb was lower. Most of the changes in the cartilage involved decreased deep binding water content in patients with no genu recurvatum but only in the tibia cartilage at the posterior horn of the medial meniscus of the affected limb in the patients with genu recurvatum. Compared with that in patients with mild genu recurvatum, the deep free water content of the cartilage in the patients with severe genu recurvatum was increased, the binding water was decreased, and the free water in the shallow layer was decreased. Additionally, in patients with severe genu recurvatum, the deep free water content of the femur cartilage at the posterior horn of the medial meniscus was lower than that of the patients with mild genu recurvatum. Therefore, we can observe a trend: free water in the shallow and deep layers of the cartilage first increased and then decreased in the patients with no genu recurvatum, the patients with mild genu recurvatum and the patients with severe genu recurvatum, while the deep cartilage binding water continued to decrease in the patients with mild genu recurvatum and the patients with severe genu recurvatum. This trend was very similar to the trends in the degree of osteoarthropathy progression. At the early stage of cartilage injury, matrix metabolism is very active, matrix synthesis increases, and proteoglycan content is increased. Afterwards, the collagen grid is damaged, proteoglycan is lost, and the free water content is increased 41 . Hence, the knee joint changes caused by the degree of genu recurvatum are closely related Figure 1. Histograms display statistically significant knee cartilages between hemiplegic patients and healthy volunteers, in which one or more of the four indices had p < 0.05 (*). In the patients with left limb hemiplegia, the hemiplegic limbs (A) and the healthy limbs (B) were compared with the corresponding limbs of the healthy volunteers. The hemiplegic limbs (C) and the healthy limbs (D) in patients with right limb hemiplegia were compared with the corresponding limbs of the healthy volunteers. a, The femur cartilage at the posterior horn of the medial meniscus; b, the femur cartilage at the posterior horn of the lateral meniscus; c, the tibia cartilage at the anterior horn of the medial meniscus; d, the tibia cartilage at the posterior horn of the medial meniscus; e, the thickest bearing cartilage of the lateral tibial condyle; f, the thickest bearing cartilage of the medial femoral condyle; g, the femur cartilage at the anterior horn of the medial meniscus; h, the tibia cartilage at the anterior horn of the lateral meniscus.  This study had the following shortcomings: (1) the sample size was small, and large sample tests are needed for further verification; (2) signal strength conversion was adopted instead of the real data on free water and combined water, and the next step could involve verification by UTE T2* mapping; (3) sometimes the cartilage was  www.nature.com/scientificreports/ thin, inevitably yielding some errors when measuring the deep and shallow layers; (4) there were no data on any mechanical changes in the knee joint with genu recurvatum, and these changes could not be perfectly explained.

Patient population. This study was approved by the Ethics Committee of the Beijing Rehabilitation
Hospital (Protocol Number: 2020bkkyLW008). All methods were carried out in accordance with the relevant guidelines and regulations. Patients were hospitalized in the Neurological Rehabilitation Center of the Beijing Rehabilitation Hospital from January 2019 to October 2020, and healthy volunteers included staff members and students. All the patients and healthy volunteers in the study provided written informed consent. Inclusion criteria: (1) initial onset of stroke, course of disease lasting more than 6 months, with hemiplegia on one side of the limb; (2) no previous history of knee injury; (3) ability to walk independently and safely at least 100 m and having walked for at least 3 months before inclusion, with a FAC grade III or above; (4) Brunnstrom stage III and above of the hemiplegic lower extremity; and (5) no obvious abnormal signals in the knee cartilages and meniscus on conventional T2W MRI and PD imaging.
The loss of normal cartilage volume begins around the age of 40 years 42 , hence, healthy patients aged approximately 40 years were selected as the control group instead of age-matched healthy patients. Inclusion criteria for healthy volunteers: (1) age between 35 and 50 years old; (2)  Imaging processing and data measurement. After acquisition, the data were transmitted to a GE AW4.6 workstation. The images were processed and measured by a radiologist with 16 years of experience in MR image processing. After selecting a patient or healthy volunteer, the TE = 0 images were first opened in the 3D synchroview for selection of the appropriate sagittal level in which the measured cartilage was the best displayed. The signal intensity of the cartilage at the thickest load-bearing point of the medial and lateral femoral condyles, the thickest load-bearing point of the medial and lateral tibial condyles, the femur cartilage at the anterior and posterior horn of the medial and lateral meniscus, and the tibia cartilage at the anterior and posterior horn of the medial and lateral meniscus were measured. The high signal intensity line of the calcified layer was used as the dividing line between the cartilage and subchondral bone. The thickness of the whole cartilage was measured first, and then the cartilage was divided into a shallow layer and a deep layer according to the Williams 43 method (Fig. 5), with the centre of the cartilage thickness as the boundary. The shallow layer spanned the centre of the cartilage thickness to the articular surface, and the deep layer spanned the centre of the cartilage thickness to the high signal intensity line of the calcified layer. To avoid errors, an elliptical region of interest (ROI) was used. The ROI size was defined to cover the entire cartilage area to be measured. During the measurement of each position, www.nature.com/scientificreports/ the centre of the cartilage length was selected as the measurement area; the position was measured once before and after, and the value of the centre position was averaged as the final measurement value. The tibia cartilage at the anterior and posterior horns of the medial and lateral meniscus were measured differently. The cartilage here is very thin and difficult to measure. The length of half of the cartilage near the centre was taken for measurement, and the centre of the cartilage was used as the measurement point. Then, the TE = 4.6 ms images were opened and the above measurement process was repeated.
In the TE = 0 images, the signal intensity of the cartilage is mainly formed by the hydrogen protons of free water and bound water. In the TE = 4.6 ms images, the signal intensity of the cartilage is mainly formed by the hydrogen protons in free water. Therefore, the relative content of free water in the shallow and deep layers of the cartilage can be calculated according to the signal intensity, that is, the percentage of free water content = I (TE = 4.6 ms)/I (TE = 0 ms) × 100%, where I (TE = 4.6 ms) and I (TE = 0 ms) are the signal intensities in the TE = 4.6 ms and TE = 0 ms images, respectively, and then the ratio of free water in deep and shallow layers and the ratio of bound water could be calculated. Statistical analysis. Statistical analysis was performed using SPSS 17.0 statistical software. The cartilage at the 12 positions, including the thickest load-bearing cartilages of the medial and lateral femoral condyles, the thickest load-bearing cartilages of the medial and lateral tibial condyles, the femur cartilages at the anterior and posterior horn of the medial meniscus and lateral meniscus, and the tibia cartilages at the anterior and posterior horn of the medial meniscus and lateral meniscus, were compared between stroke patients with hemiplegia and healthy volunteers, between the healthy limb and paralyzed limb in stroke patients, and among stroke patients with no genu recurvatum and mild and severe genu recurvatum. The mean ± standard deviation is used to represent the data. The independent sample t test was used for all mean comparisons. When the data did not conform to a normal distribution or lacked homogeneity of variance, the Mann-Whitney test was used. P < 0.05 was considered statistically significant.

Conclusion
Genu recurvatum in stroke patients with hemiplegia could cause changes in the moisture content of the knee cartilage. As the severity of the genu recurvatum increases, the changes in knee cartilage become more obvious. Both the affected limb and the healthy limb in stroke patients with hemiplegia exhibited corresponding changes. The so-called "healthy limb" may no longer be considered "healthy" and could not be used as a normal control; rather, it should be considered simultaneously with the affected limb in the development of a rehabilitation treatment plan.