The amputation of a limb is a surgical procedure that is regularly performed in small animal practice. Severe trauma and limb tumors are the most common reasons for performing an amputation; other indications include chronic osteomyelitis, neurological dysfunctions such as sciatic neuropathy and brachial plexus paralysis, congenital limb deformities, vascular disease and arteriovenous fistulas [1–3].

In spite of several clinical reports indicating high owner satisfaction after limb amputation in dogs [4–7], an amputation is still very critically seen by the owners, and even by some veterinarians. Particularly, owners have the tendency to think that the procedure may affect the animals emotionally, as it indeed happens in people [8], or that it will be disabling for them. Besides, owners are often worried about the possibility of overload of the remaining limbs, leading to hypothetical secondary joint pathologies. Thus, many owners are reluctant to have their dog amputated and reject the amputation as an alternative to euthanasia or take the decision only after the patient has gone through a painful surgical and/or medical treatment process.

The lack of objective information prevents the veterinarian from providing the owners with accurate information about these concerns. A hesitant veterinarian might then play a role in the owner deciding against the amputation. There is only one previously published report objectively evaluating the gait of amputated dogs [9]. In that study, force plate analyses were carried out to measure ground reaction forces (GRF) and contact times in a population of 10 large-breed dogs which had a limb amputation (five forelimbs and five hind limbs). Additionally, the center of gravity was calculated in those patients. It was found that the absence of a limb caused statistically significant changes in the GRF, impulses and contact times of the remaining limbs and the location of the animal’s centre of gravity, in comparison to a control group of 22 healthy dogs. However, there are no prospective studies with animals which are planned to be amputated, and no study has been performed objectively evaluating kinematics (joint movement) or possible joint changes after a hind limb amputation in dogs.

The general objective of the present study was therefore to prospectively characterize the recovery outcome of dogs undergoing a hind limb amputation. In order to evaluate the motion and weight bearing characteristics, as well as the duration of adaptation to the three-legged gait, kinematic and kinetic analyses were carried out. Furthermore, MR images of the remaining contralateral femorotibial (stifle) joint were made before and 4 months after the amputation, in order to investigate possible changes in joint morphology, due to a hypothetic weight bearing overload of this limb.

It was hypothesized that there would be marked changes both in the kinetic and the kinematic parameters after the amputation, but that those changes would not impair the ability of the animal to lead a normal life. Based on our clinical experience and some of the aforementioned studies [4–7], it was also hypothesized that there would not be any changes in the contralateral stifle on the MR examination. Thus, after the initial reluctance to the amputation, owners would be satisfied with the procedure.

This study was carried out in accordance with the German Animal Welfare Guidelines and was approved by the Ethics Committee of the Lower Saxony State Office for Consumer Protection and Food Safety (Approval Number: 10A071). All owners agreed to their dogs participation in the study and signed a consent form.

Kinetic and kinematic gait evaluation

Kinetic (forces) and kinematic (movement) gait analysis was performed one to three days before the amputation, as well as 10, 30, 90 and 120 days after surgery.

Kinetics were measured using a specially designed treadmill¹² consisting of four separate belts, each of them with an integrated force plate underneath. This design allowed the simultaneous measurement of all limb forces.

Kinematic analysis was performed with the aid of retro-reflective markers (Ø 16 mm reflective markers)¹³ positioned on 24 anatomic landmarks (8 per remaining limb), using double-sided adhesive tape; the location of these markers has been previously described [10, 11] and is illustrated in Fig. 1. Six high-speed infrared cameras¹⁴ were used to record marker movement in all three remaining limbs simultaneously, as the animals were walking at a controlled speed (measurement frequency: 100 Hz). Before each measurement, static and dynamic camera calibration was performed using an L-shaped calibration device.¹⁵

On each gait analysis session, patients were gently introduced to the gait on the treadmill; on the first day, a speed at which each individual patient walked comfortably on the treadmill was determined; on each subsequent session the patient was evaluated using the same speed, ranging from 0.5 to 0.8 m/s. During each gait analysis session, two to six trials were recorded, each with a duration of approximately 30 s, until at least one valid trial was obtained. A valid trial was defined as 10 consecutive regular steps, in which the dog walked smoothly, without any external forces from the handler being applied, with all paws landing on the appropriate force plate, without overstepping. Video recording was performed, to ensure that the steps were appropriate for analysis.

Both kinetic and kinematic data were simultaneously recorded using commercially available software.¹⁶

Ten consecutive steps were afterwards analyzed for the following kinetic parameters: peak vertical force (PFz), mean vertical force (MFz), and vertical impulse (integral [IFz]). All forces were normalized to the individual body weight of each dog and data were expressed as percentage of body weight (%BW). Mean ± standard deviation (SD) was calculated from 10 valid consecutive steps. Afterwards, load redistribution (LR) was calculated for each forelimb and remaining hind limb, for each measured parameter (PFz, MFz, IFz), using the following equation (according to Steiss et al. [12]): % load bearing = Fz of the limb/total Fz of all limbs*100. Other kinetic parameters were not calculated to avoid overloading this section with too much information. The kinetic data were processed using commercial software¹⁷ and exported to a commercially available spreadsheet.¹⁸

In order to process the kinematic data in Vicon Nexus, all markers were labeled in a trial. Then, 10 valid foot strikes were marked manually to define the gait cycle (stance and swing phases) of each limb. Using a 2-dimensional (2-D) model, projected flexion and extension angles of each remaining joint were calculated: contralateral (with respect to the amputated hind limb) scapulohumeral joint, contralateral cubital joint, contralateral carpal joint, ipsilateral (with respect to the amputated hind limb) scapulohumeral joint, ipsilateral cubital joint, ipsilateral carpal joint, contralateral coxofemoral joint, contralateral femorotibial joint and contralateral tarsal joint. Measured angles are illustrated in Fig. 1. In order to compare the movement pattern of each analyzed joint, the gait cycles were normalized to 100 in all dogs and displayed as percentage of one whole stride. The mean joint angle (±SD) and the range of motion (±SD) of the aforementioned joints were calculated from the mean joint angle progression curves (MJAPC) calculated from the 10 strides per dog. Mean joint angles and range of motion were used since the reader can easily understand their comparison between sessions and to avoid overloading the manuscript with too much data. The kinematic data were processed using commercial software¹⁹ and then exported to a commercially available spreadsheet¹⁸.

The main goal of this study was to provide owners and veterinarians with accurate, objective information, not existing previously, about the outcome of the amputated patient (i.e. how the dog adapted to the new locomotory status).

Examination times were chosen using the data of a previous study [7] which indicated that most dogs adapted to the amputation within a month, some within a week, and all within 3 months after surgery. With the last follow-up 120 days after the procedure, a final attempt was made to find any evidence of further gait changes. Since there is the belief that orthopedic disease might occur in the remaining limb after amputation, as a result of a theoretical overload, the contralateral stifle was morphologically assessed using MR images, to evaluate this likelihood. The femorotibial (stifle) joint was chosen, instead of the coxofemoral or tarsal joint, as the former has been extensively investigated, and there is a whole body of literature exemplifying normal and pathologic MR images to compare [14–19].

As expected, a very heterogeneous population was found in this study: different breeds, ages and weights. The main reason for amputating a patient was the presence of a tumor, which agrees with previous studies [6, 7]. The mortality seen in this study was not related the amputation itself.

The kinetic and kinematic analyses revealed that the patients had begun adapting to the new locomotory situation even before the amputation was performed, and that 10 days after the procedure, all major changes had already taken place. It was unfortunate that it was not possible to perform all examinations on all patients. However, exclusively using “good” trials allowed us to obtain reliable results. It should be noted that, even though a valid trial implied that the handler did not exert any external forces, it was not possible to have the animals completely free on the treadmill (see Additional file 1: Video 1). Unfortunately, the willingness to walk or trot or overall activity of the patient were not evaluated in the present study; these factors would have brought additional information about the outcome. It is remarkable that the largest patient (patient 11–54 kg) and the smallest one (patient 10 – 8 kg) adapted equally well. These results agree with the study of Kirpensteijn et al. [7], indicating that, even though subjective, the observations made by the owners of such study were actually very precise. The present study has the advantage of looking at the patients objectively and prospectively, leaving no doubt about the fast adaptation of all animals. The fact that there were no significant changes in the load redistribution after amputation was initially surprising; however, it is easily explainable: all patients were severely lame before the amputation, meaning that the adaptation and compensation to the lack of a hind limb had begun to take place before the amputation was performed. This was even clearer when looking only at those patients which did not load the affected limb at all before surgery. It should also be kept in mind, that the lack of statistically significant changes could have resulted from the low number of dogs included in this study, and therefore in the lack of power of the tests.

The kinematic results (including the statistical analysis) should be interpreted with caution. The fact that significant differences were found in the absolute values of the different joint angles could be explained by the normal variation inherent to motion analysis: marker localization changes lead to changes in joint angles measured. Even though every effort was made to place the markers in the right position, small variations might have occurred, leading to different measurements. However, also huge variations were seen within a session (not shown in Table 3 or Figs. 3, 4 and 5), possibly indicating that the patients adapted to every step they made and in a very irregular manner. Affected balance of the dog during movement and the resulting instability might cause these variations. Another possible explanation is the fact that several gait patterns are possible at a given speed and by a permanent adjustment of the speed of the animal to the treadmill speed [20]. It should also be taken into account that slow speeds were used before the amputation, in order to avoid worsening the pain that patients were already experiencing. After the amputation the animals could have walked on the treadmill more comfortably at faster speeds (personal observations). Even though this might have caused an irregular walking pattern after the procedure, speeds were kept constant to avoid adding a variance factor when measuring the GRF [21]. Finally, the angles measured here can be used to illustrate the movement patterns for the patients in this study, and they cannot be extrapolated to other patient populations. However, our study focuses on determining whether there are variations in the different kinematic parameters before and after amputation in this particular patient population, and that does not seem to be the case.

The similarity in the ROMs before and after surgery is a remarkable finding. The measurement of ROMs seems to be less susceptible to the sources of error commonly found in kinematic studies (misplacement of markers and skin movement), at least in the hind limb [22]. Therefore, the results of the ROMs are more accurate than the absolute measurement of joint angles. That being said, the lack of significant differences between sessions suggests that the patient had also begun to adapt to the new movement situation before amputation, and that this remained stable after the procedure. However, it should also be taken into consideration that joint angular curves in each measurement points are very similar in shape, resulting in stable ROMs; this could also explain why the ROMs showed much less variation than mean joint angles.

The lack of changes in the MR examination of the remaining contralateral stifle after amputation could indicate that an overload in the remaining contralateral hind limb, leading to joint pathology, is not very likely. It was decided to investigate this point, as it is commonly believed by the general public, and even by some veterinarians, that amputating a dog might predispose it to orthopedic abnormalities; the results of our MR (and also our physical examinations) proved otherwise, at least for the stifle. Although 4 months after amputation might possibly be a short time to evaluate joint changes, a previous study describing the experimentally induced rupture of the CrCL in 5 crossbred dogs showed that it is possible to see changes in the cartilage and subchondral bone as early as 4 weeks after the rupture [17]. It is of course difficult to extrapolate such findings to this study, but it could indicate that, if there were ongoing changes in joint morphology, they would be visible 4 months after amputation. Besides, most of our research subjects were oncologic patients, with a (likely) short life span. Measuring cartilage thickness would have been a more accurate method to evaluate subtle joint changes [17]. Unfortunately, this could not be done due to software limitations. In any case, we did not expect to see any changes, as our clinical experience indicates that no changes are seen in the contralateral limb after a hindlimb amputation; nevertheless, it is not known what kind of changes an amputation would cause in patients suffering from degenerative joint disease in the remaining limbs.

With regard to the owner evaluation assessment of the patient after the procedure, the results were as expected: in Table 4, the positive outcome of most patients can be clearly seen. The improvement is especially remarkable in the ability of the dog to lie down and stand up. The lack of improvement or worsening of some parameters for some patients seemed to be related more to the declining general condition of the patient, than to the effects of the amputation itself. The use of an objective tool to measure patient activity, such as an accelerometer, could have provided more accurate information about the outcome and they have been previously used in animals [23]; however, this tool was not available to us.

The responses to the final questionnaire were also as expected: most owners were initially reluctant to have their dogs amputated, but were satisfied with the overall result and the quality of life was considered good; this is in agreement with other studies [4–7]. In the present study, some owners even considered that their pet’s quality of life improved after the amputation, and this might have been related to the removal of the source of pain. The behavioral changes reported by the owners in this study were more small disabilities than behavioral problems. The behavioral problems previously reported [7], such as aggression and anxiety, were not seen in the patients of the present study.

As in previous studies [4, 7], all owners responded that they would have another pet amputated, and none regretted the decision. We believe that an evaluation made by owners whose pets died shortly after the procedure would have been negatively biased and they were not gathered.

The favorable responses of most owners can be explained by the fact that the dogs adapted very soon to their new locomotory situation (kinetic and kinematic analyses) and to the lack of morphologic changes in other joints (as it might be inferred from the lack of morphologic changes in the stifle).

In spite of the limitations, this study provides objective evidence indicating that dogs have a quick adaptation process after a hind limb amputation. The adaptative processes to the new locomotion begin even before the amputation is performed. Since the veterinarian is responsible for providing accurate information before an amputation [2], we strongly believe that this study provides useful information, that will allow veterinarians the possibility to give dog owners more realistic expectations of a hind limb amputation.