CORR Synthesis: What Is the Evidence for Age-appropriate Alignment Goals in Surgery for Adult Spinal Deformity?

In the Beginning…

The transition to patient-specific care in orthopaedic surgery encompasses an overarching effort to customize treatment plans based on each patient’s genetics, biomarkers, and a unique signature defined by imaging and other diagnostic modalities. The goal is to maximize the ability to predict a priori which patients will benefit from surgery and to individualize procedures to the extent possible.

Within the field of adult spinal deformity surgery, multiple classification algorithms exist to categorize global malalignment based on variations in spinal morphology, severity of deformity, and age. The initial breakthrough in our understanding of patterns of variation in sagittal alignment began with Pierre Roussouly's work with Johanes Dimnet, a biomechanical engineer, 15 years ago (Written communication, P. Roussouly, MD, January 16, 2020). They proposed a geometric model consisting of two arcs of a circle, with the tangent on the apex. The lower limit was the sacral plateau and the upper limit was free at the point where lordosis transitions into kyphosis. In this model, the lower arc was equal to sacral slope, closely linking lordosis and sacral slope. By introducing the variation of sacral slope inherent to the normal population, it was possible to predict distinct classes of lordosis curvature by lengthening or shortening the lower arc. This supplied the basis for the 2005 seminal paper by Roussouly et al. [27]. In Roussouly’s words when contacted for this review, “It took probably more than 1 year between the geometrical model and its application for classification” (Written communication, P. Roussouly, MD, January 16, 2020). This simple yet novel mathematical model was the catalyst for the discovery process and the concept that there is no single paradigm of ideal global alignment.

The Argument

Substantial evidence supports a correlation between sagittal alignment and patient function and patient-reported outcomes [9, 10]. Recent work highlights the importance of accounting for these age-related changes in spinopelvic alignment, and the effect of these changes on surgical realignment goals [13, 23]. To date, however, only a relatively low percentage of patients with adult spinal deformity, approximately 20.6% to 33.3%, achieve age-appropriate alignment thresholds after surgery for adult spinal deformity [14, 23, 24, 28]. The sagittal vertical axis (SVA) increases with age, in association with increasing thoracic kyphosis and pelvic tilt. Overcorrection of sagittal alignment in patients older than 65 years may result in an increased risk of mechanical and neurological complications [12, 18, 25].

With improvements in preoperative medical planning and the development of less invasive surgical techniques, more older patients are candidates for spinal deformity surgery [1]. Such patients are likely to be seen in all spine surgeons’ practices, regardless of whether or not they perform a high volume of spinal deformity surgery.

Essential Elements

To identify all original prospective and retrospective studies for this review, we searched PubMed, Embase and abstracts derived from the Scoliosis Research Society and International Meeting on Advanced Techniques Annual Meetings related to age-specific alignment goals. We searched the terms “adult spinal deformity”, “age”, and “alignment” in PubMed and Embase, which yielded 693 results. After removing 387 duplicate articles, we conducted an abstract review of 306 studies. We initially excluded 16 descriptive review articles and 62 basic science/biomechanical studies. We then excluded 213 studies unrelated to the topic, or lacking assessment of radiographic and/or health-related quality of life data, and descriptions of complications. This ultimately left 15 studies to serve as the basis for our report.

As this topic has only come to the forefront in the very recent past, there are no prospective or randomized controlled studies evaluating surgical groups for whom age-appropriate alignment was considered and those for whom it was not. Therefore, this review is limited to observational research only; five were single-center studies, and 10 were multi-institution studies. The quality of the included studies was assessed using a “star system” following the guidelines of the Newcastle-Ottawa Scale, judging the selection of the study groups, the comparability of the groups, and the ascertainment of the outcome of interest [30] with a possibility of nine total stars. All studies were deemed to be of good quality, receiving three or four stars in selection, one or two stars in comparability, and two or three stars in outcomes.

What We (Think) We Know

The process of aging encompasses diverse physiologic changes including fatty infiltration of muscle and time-dependent decreases in skeletal muscle mass [7, 29]. Postural characteristics of older patients include forward positioning of the spine and compensatory mechanisms including retroversion of the pelvis, hip extension, and knee flexion until strain or muscle atrophy render these mechanisms impossible and the patient begins to rely on assistive devices for ambulation. Conversely, degenerative changes in the knee that limit extension, thereby leading to increases in knee flexion or knee flexion contractures, may cause compensatory decreased lumbar lordosis in older patients [20]. Postural control deficits in an older patient may also be the result of cognitive decline or an impairment in visuomotor performance, which is the ability to synchronize visual input with physical body movements [4].

For years, we have known that loss of lumbar lordosis drives sagittal malalignment [5]. As lumbar degeneration and thoracolumbar coronal deformity lead to changes in lumbar lordosis over time, our quantitative analyses have come to largely rely on pelvic incidence. Patients can be classified as having either low or high pelvic incidence. During realignment surgery, patients with a high pelvic incidence need the apex of their lumbar lordosis placed more cephalad than in those with a low pelvic incidence. Creating apical lumbar lordosis too caudally in patients with a high pelvic incidence may result in higher rates of rod fracture and proximal junctional kyphosis [6]. This concept translates into an important technical consideration in planning the location of realignment lumbar osteotomies in older patients undergoing adult spinal deformity surgery [15].

Given the baseline they start from, older patients likely tolerate a higher degree of residual sagittal malalignment postoperatively. A study of 773 patients demonstrated that patients older than 65 years displayed substantially greater sagittal malalignment than younger patients, but despite this greater deformity level, they had equivalent patient-reported health outcomes [13]. In this study, patients 65 to 74 years of age displayed a moderate level of disability, defined as an Oswestry Disability Index (ODI) equal to 20, with a pelvic incidence lumbar lordosis mismatch (PI-LL) of 5.5° and a sagittal vertical axis (SVA) of 50.4°, compared with patients who were 45 to 54 years old with equivalent disability (ODI = 20), and a much lower PI-LL of 0.2°, and SVA of 21.6°. Even more striking was that patients older than 74 years of age demonstrated the same level of disability with a PI-LL of 8.3° and SVA of 65.8°, values that denote significant sagittal malalignment. This study laid the framework for the concept that older patients tolerate a higher degree of sagittal malalignment than younger patients and provided concrete values for realignment parameters in patients with adult spinal deformity of varying ages [13]. Similarly, a multicenter study reported that in adult spinal deformity surgery incorporating lateral lumbar interbody fusion or minimally invasive transforaminal lumbar interbody fusion coupled with either open or percutaneous pedicle screws, 67% of patients older than 65 years had evidence of postoperative sagittal malalignment, defined as a PI-LL mismatch greater than 10° or SVA greater than 50 mm [23]. However, health-related quality-of-life outcomes were not different between older patients who exhibited postoperative sagittal malalignment and those who did not; both groups improved substantially with surgery.

To assess preoperative and postoperative global alignment accurately, radiographic evaluation should include an analysis of the lower extremities, specifically the hip and knee posture, in addition to spinopelvic parameters. Although radiographic deterioration may be the hallmark of undercorrection or overcorrection after adult spinal deformity surgery, persistent postoperative recruitment of lower-extremity compensatory measures may reflect inadequate restoration of the ideal sagittal alignment for a patient [29]. A recent multicenter study of 108 patients found that younger patients (40 to 65 years old) who were undercorrected based on age-adjusted targets, demonstrated greater postoperative recruitment of lower-extremity compensatory mechanisms including pelvic retroversion and knee flexion [24]. Interestingly, patients older than 65 years whose surgical correction attained age-adjusted alignment targets demonstrated the greatest amount of lower extremity compensation postoperatively, reflecting the inherent and progressive flexion tendency of the knee in stance as a patient ages, irrespective of surgical modifications to spinal sagittal alignment. This underscores the concept that in the older patient, lower-limb compensatory mechanisms will continue to increase naturally with age, an important consideration for planning the extent of spinal sagittal plane correction.

Overcorrection of sagittal malalignment can result in mechanical complications with potentially devastating neurologic sequelae. An analysis of 679 patients with adult spinal deformity demonstrated a higher rate of proximal junctional kyphosis in patients who were overcorrected, defined by a smaller pelvic incidence lumbar lordosis mismatch and sagittal vertical axis, than their recommended age-specific targets [14]. Similarly, a follow-up study of 625 patients with adult spinal deformity demonstrated a 9.7% higher rate of proximal junctional failure in patients who had overcorrection of sagittal alignment after surgery compared with age-specific alignment targets, without the incorporation of surgical techniques designed to mitigate the risk of proximal junctional failure such as proximal hooks, tether constructs, or upper instrumented level vertebroplasty [17]. Importantly, to our knowledge, no studies have used motion analysis in adult spinal deformity surgical populations of different ages and alignment targets, to determine differences in forces transmitted at the transitional junctions. Although it is clear that overcorrection of sagittal malalignment relative to age-specific targets is a potent risk factor for proximal junctional kyphosis (PJK) and proximal junctional failure (PJF), it is one of many factors that play a role in the development of these catastrophic complications.

Although minimizing mechanical complications such as PJF is a driving force to incorporate age-specific alignment parameters in preoperative planning, the importance of achieving age-specific alignment also lies in its utility as a surrogate measure for improved postoperative health-related quality of life (HRQOL) data. A retrospective multicenter analysis of 343 patients demonstrated that patients who had undercorrection based on their age-adjusted parameters had worse ODI, SF-36 physical component score (PCS), and Scoliosis Research Society (SRS) scores at 2 years [28].

Knowledge Gaps and Unsupported Practices

We know it is crucial to determine age-appropriate sagittal realignment parameters for each patient during preoperative planning, and concrete values for alignment parameters are now available and validated as ideal age-adjusted targets for age-specific ODI US-norms and age-specific SF-36 PCS US-norms [13]. One question regarding realignment targets remains: Should we plan preoperatively for realignment targets that are exactly commensurate with the patient’s current age or for their age in 5 or 10 years, as we know he or she will continue to have age-related changes in their sagittal and spinopelvic alignment [5]? This has been a challenging question to answer, since it is clear from studies that while overcorrection imparts minimal to no benefit, undercorrection of the deformity has an adverse impact on postoperative HRQOL data [28]. Therefore, most deformity surgeons’ inclination is to ensure correction meets age-adjusted alignment targets based on the patient’s age at the time of surgery. Further, all patients with adult spinal deformity demonstrate some measure of postoperative reciprocal and compensatory alignment and postural changes over time, depending on patient variability in flexibility of the unfused segments, muscle tone, and muscle fatigue [2, 22]. We are still in the early stages of understanding loss of correction and reciprocal changes over time. In this regard, adult spinal deformity bears resemblance to a chronic disease process, and surgery is not by any means a static “cure”. Long-term (5- and 10-year follow-up) analyses of global alignment after adult spinal deformity surgery, HRQOL data, and rates of mechanical failure will help answer the question of whether to adhere to age-adjusted targets at the time of surgery or incorporate the effects of time.

Barriers and How to Overcome Them

Scores such as the Global Alignment and Proportion score quantify overall spinopelvic alignment, with age taken into consideration, to allow surgeons to tailor specific realignment goals for patients with adult spinal deformity. In this score, as in other classification systems, age is used as a surrogate for underlying pathophysiologic processes such as osteoporosis, sarcopenia, and deterioration of musculoskeletal function, which result in deficits in balance and stability control. However, age is only a rough measure of this process, and the age-related trajectory of musculoskeletal decline may vary greatly depending on variables such as lifestyle, smoking history, comorbidities, and physical activity level. More accurate biomarkers of aging could potentially include motor function tests that assess hand grip strength, standing balance, and endurance, among other variables [11, 26]. Urine and serum biomarkers of skeletal muscle mass, bone turnover, and cartilage degradation are also candidate markers of the aging process and may have a greater ability to predict instrumentation failure and mechanical complications than age alone [8].

The studies described in this review have, for the most part, assessed only radiographic parameters of sagittal alignment and HRQOL data as endpoints at 1 or 2 years of follow-up. None of the studies include data on the rate of complications and revision procedures within cohorts that were under- or overcorrected based on age, and the effect these complications and unplanned reoperations may have on patient-reported outcomes at long-term follow-up. A granular analysis of the frequency of revision surgery in patients specifically related to undercorrection or overcorrection based on age-specific alignment targets would shed light on how these complications contribute to decreased cost effectiveness and worsening HRQOL over time.

5-year Forecast

Spine surgeons are able to perform bony realignment, and in this regard, our research efforts are focused. We use radiographic criteria to describe current alignment, age, and other patient variables as surrogates for musculoskeletal pathophysiology and make the best predictions for outcomes after reconstructive surgery. However, the concepts of sagittal alignment and global balance are multifactorial and incorporate neurosensorial modulation, integration of proprioceptive and vestibular systems, and intrinsic patient characteristics such as skeletal muscle and bone mass.

Current systems to derive age-appropriate realignment targets do not consider bone mineral density, frailty, and BMI, which are potent predictors in many large studies of postoperative complications in patients with adult spinal deformity [3, 16, 21, 31]. Given the depth and complexity of data inputs, predictive modeling such as supervised machine learning has the greatest promise, with appropriate feature reduction to include only the most informative features relevant to complications. Further work is also required to understand how a balanced preoperative team approach addressing these factors can help prepare older patients for surgery and improve overall care. These factors described above might also be included in a point-of-care assessment of the patient’s individual risk for complications and prolonged length of stay after adult spinal deformity surgery. Progress toward this goal could involve standardized incorporation of a frailty index calculation in the electronic medical record system, generating a value for each patient at the time of initial consultation or surgical discussion, as well as corresponding odds ratios for specific types of complications [19]. We believe large dataset and multicenter validation of the predictive accuracy of such models will augment the expertise of the surgical team treating patients with adult spinal deformity.