Adult spinal deformity (ASD) is a debilitating condition associated with substantial pain and disability, ultimately reducing patients’ health-related quality of life (HRQOL) [1]. Because positive sagittal imbalance is strongly associated with poor HRQOL [2,3], optimal restoration of spinopelvic alignment has become a cornerstone of surgical management for ASD to achieve favorable clinical outcomes, including maximized function and minimized complications [4–7]. One major complication associated with suboptimal sagittal-plane correction is proximal junctional kyphosis (PJK), which may progress to its more severe form, proximal junctional failure (PJF). PJK is defined as an abnormal kyphotic angulation (e.g., a proximal junctional angle >10°) at the upper end of a fusion construct, whereas PJF indicates structural failure at the junction that necessitates revision surgery.

Since the early 2000s, multiple classification systems and alignment metrics have been developed to guide surgical planning in ASD. This narrative review examines five major sagittal alignment frameworks and their relationship to patient-reported outcomes and proximal junctional complications: (1) Scoliosis Research Society (SRS)–Schwab classification, (2) age-adjusted alignment goals, (3) the Global Alignment and Proportion (GAP) score, (4) the Roussouly classification, and (5) vertebral-pelvic angle (VPA) metrics. A literature search of PubMed/Medline and major spine society publications was conducted to identify studies published between 2000 and 2024 using keywords related to ASD, sagittal alignment, HRQOL, and proximal junctional complications. Landmark cohort studies, meta-analyses, and alignment-specific validation articles were prioritized, while literature on pediatric deformity and cervical alignment literature was excluded. This review synthesizes evidence from these sources to provide a focused appraisal of contemporary sagittal alignment strategies, with particular emphasis on HRQOL outcomes and the prevention of PJK/PJF.

Schwab and colleagues defined radiographic thresholds for spinopelvic parameters associated with an Oswestry Disability Index (ODI) >40, indicating moderate disability, as pelvic incidence minus lumbar lordosis (PI–LL) >11°, pelvic tilt (PT) >22°, and sagittal vertical axis (SVA) >47 mm in non-operative patients with ASD [3]. These findings subsequently formed the basis for the SRS–Schwab classification (Fig. 1). Introduced in the 2012, the SRS–Schwab classification provides a standardized framework for categorizing ASD severity based on radiographic parameters [8]. It integrates descriptive curve types with three sagittal spinopelvic modifiers that were empirically selected for their strong associations with HRQOL outcomes. Using these modifiers, patients can be stratified as well aligned or moderately to severely malaligned. Importantly, each stepwise increase in sagittal modifier grade is associated with progressively worse HRQOL. In a landmark validation study of 527 ASD, Terran et al. [9] demonstrated that individuals with more severe SRS–Schwab modifier grades had significantly higher ODI scores and worse SRS-22 outcomes. As a result, SRS–Schwab grades can assist in guiding treatment decisions, as patients with worse sagittal grades are more likely to be surgical candidates and often require more extensive procedures, including osteotomies and pelvic fixation, due to their significant functional impairment [8–10]. Although the SRS–Schwab classification was initially developed using nonsurgical cohorts, subsequent studies have validated its applicability in surgically treated patients. Smith et al. [10] demonstrated that improvements in each sagittal modifier at 1-year postoperative follow-up were significantly associated with better HRQOL outcomes. Similarly, Lee et al. [4] reported long-term outcomes with a mean follow-up of 90 months, showing that patients in whom all three parameters were corrected to grade 0 modifiers experienced significantly greater clinical improvement than those who were in inappropriately corrected.

Despite its value in assessing deformity severity and HRQOL impact, the SRS–Schwab classification is less effective in predicting mechanical complications, including PJK/PKF [4,11–13]. This limitation may stem from its inability to fully account for individual patient characteristics such as age and PI. For instance, applying identical alignment target to an 80-year-old and a 60-year-old patient may not be clinically appropriate given differences in physiological reserve and age-related factors. Furthermore, although a PT value of ≤20° has been defined as normal, individuals with a large PI may physiologically demonstrate PT values exceeding this threshold [14–16]. Therefore, the use of a fixed PT cutoff without consideration of PI may not be suitable for all patients. Finally, unlike PI–LL, both PT and SVA are positional parameters that can only be measured in the standing position, which may limit their direct applicability during surgery. Key results are summarized in Table 1.

Age-adjusted alignment goals emerged from evidence demonstrating that ideal sagittal parameters change with age as a result of natural, senescent alterations in the spine and pelvis. Lafage et al. [17] pioneered this concept by analyzing the relationship between age and optimal spinopelvic parameters using regression analyses that incorporated age-adjusted patient-reported outcome measures derived from a normative population. In their study, the calculated optimal alignment values for patients >75 years were PT=28°, PI–LL=17°, and SVA=78 mm, compared to PT=11°, PI–LL=−10°, and SVA=4 mm for patients <35 years. The central principle of age-adjusted alignment is that sagittal alignment targets should be progressively relaxed with increasing age. In a subsequent study, the same group reported that corrections exceeding age-adjusted targets (defined as overcorrection) were associated with a higher incidence of PJK [18]. In this scheme, age-adjusted normative values for sagittal parameter were calculated using the following formulas [19]:

For each parameter, the ideal alignment range was defined as ±10 years around the patient’s age. For example, in a 75 years old patient, the optimal PI–LL target range would be between 8° and 18°. Follow-up studies demonstrated that overcorrection relative to age-adjusted PI–LL targets significantly increased the incidence of PJK, radiographic progression following PJK development, and recurrent PJK after revision surgery [20–26]. In contrast, whether undercorrection confers a protective role against PJK remains unclear, as current evidence is inconclusive. Im et al. [27] demonstrated that, compared with overcorrected patients, undercorrected patients exhibited larger postoperative SVA values and worse ODI scores, but similar rates of PJK. Conversely, Rothenfluh et al. [28] found that patients with a PI–LL mismatch greater than 10° had a tenfold increased risk of developing adjacent segment disease. Meanwhile, Byun et al. [21] and Sebaaly et al. [29] observed that undercorrection resulted in significantly lower rates of PJK compared with overcorrection.

Because age-adjusted alignment targets are less stringent than conventional sagittal parameters, concerns have been raised that the HRQOL outcomes may be compromised even in patients with matched correction. Protopsaltis et al. [24] compared “normative” versus “age-appropriate” targets in ASD surgery and found that the aggressively corrected patients (those targeting normative values) did not demonstrate superior HRQOL outcomes, including ODI and SRS-22, despite achieving more “perfect” radiographs compared with the age-appropriate group. Similarly, Scheer et al. [30] found no significant differences in 2-year clinical improvement between matched and overcorrected patients, with overcorrection being an established risk factor for PJK. Park et al. [31] also found that clinical outcomes, including ODI and SRS-22 scores, were most favorable in patients with matched correction relative to the age-adjusted PI–LL targets. Notably, this trend persisted even after excluding patients with PJK/PJF, suggesting that overcorrection did not provide additional clinical benefits beyond matched correction [31]. Collectively, these findings support the principle of targeting functional alignment proportional to a patient’s age and physiological capacity rather than applying uniform alignment goals.

Building on this concept, Lafage et al. [32] developed the Sagittal Age-Adjusted Score (SAAS) to quantify how closely postoperative alignment matches age-adjusted targets (Fig. 2). The SAAS assigns points based on deviations in PI–LL, PT, and T1 pelvic angle (T1PA) from established age-adjusted norms. From each parameter, the offset from the age-adjusted target is calculated and scored accordingly. In the original study, higher SAAS values were significantly associated with PJK development, with mean scores of 0.2 in patients without PJK and 4.0 in those who developed PJK. Patients with matched SAAS values also showed the most favorable HRQOL outcomes, as measured by ODI and SRS-22 scores. The findings support the concept of an optimal “sweet spot” in sagittal alignment, whereby achieving age-adjusted targets maximizes HRQOL improvements while minimizing mechanical complications. The SAAS has been proposed as a more comprehensive predictor of PJK/PJF than age-adjusted PI–LL alone, as it incorporates multiple parameters reflecting global sagittal alignment. However, subsequent studies reported that its predictive performance for PJK was not superior to that of PI–LL alone, suggesting that the predictive value of SAAS may be largely driven by the PI–LL component [33]. Furthermore, Dial et al. [34] found that the utility of age-adjusted targets for predicting mechanical complications, reporting poor discriminatory ability for deviations in PI–LL, PT, and T1PA, with C-statistics of 0.52, 0.54, and 0.48, respectively. Key results are summarized in Table 1.

The GAP score was developed as a proportional, patient-specific approach to assessing spinopelvic alignment, moving beyond fixed, absolute alignment thresholds. The system incorporates PI-adjusted measures, including relative pelvic version, relative lumbar lordosis, lordosis distribution index (LDI), relative spinopelvic alignment, and an age factor, with its score ranging from 0 to 13 points (Fig. 3) [35]. Initial results were highly promising. In the validation cohort, patients with proportionate spinopelvic alignment according to GAP experienced a mechanical complication rate of only 6%, whereas those with moderately or severely disproportioned alignment had complication rates of 47% and 95%, respectively. The GAP score demonstrated excellent discrimination ability for predicting mechanical complications, with an area under the receiver operating characteristic curve (AUC) of 0.92. Subsequent studies further supported the GAP score as a useful prognostic tool. Jacobs et al. [36] compared the predictive performance of the GAP score with that of the SRS–Schwab classification and found that the GAP score was significantly more predictive of radiographic mechanical failure following ASD surgery. Additional validation was provided by a large independent series by Gupta et al. [37], who analyzed 322 ASD patients undergoing long fusions constructs (mean 5+ year follow-up) to evaluate GAP score performance outside the original development group. Their findings demonstrated a clear association between increasing GAP categories and higher rates of mechanical complications: 21.8% in patients with proportionate alignment, 55.1% in the moderately disproportioned alignment, and 70.4% in the severely disproportioned group. Notably, patients with moderate or severe GAP scores had a 2.5-fold and 3.2-fold increased risk of mechanical failure, respectively, compared with those in the proportionate group. Moreover, a recent meta-analysis confirmed a significant relationship between GAP categories and mechanical outcomes across multiple studies [38]. In that review of 11 studies (>1,600 patients), proportionate alignment group was associated with significantly lower odds of mechanical complications compared with any degree of disproportionate alignment (pooled odds ratio [OR], 2.8 for disproportionate vs. proportionate alignment).

Despite the supportive evidence noted above, several studies have challenged the predictive power of the GAP score, reporting more equivocal or negative findings. A high-profile external validation by Kwan et al. [39] analyzed 159 patients with ASD from a multicenter prospective cohort. At 2-year follow-up, higher GAP scores were not significantly associated with an increased risk of PJK/PJF (AUC=0.60, indicating only “low” predictive ability). Likewise, the GAP score did not reliably predict revision surgery for proximal junctional complications (AUC=0.66). These conflicting findings may be attributable to differences in study designs of the original GAP investigation. The original study did not exclusively include cases with fixation extending to the sacrum or pelvis, potentially underestimating the true incidence of PJK. In addition, the primary outcome included all mechanical complications, such as rod fractures, rather than focusing solely on PJK/PJF, which may have limited its ability to predicting assessing specific risks. Yagi et al. [40] examined the GAP score in an Asian patient population and found it lacked predictive value for mechanical failure. In their multicenter Japanese cohort, the incidence of PJK/PJF did not correlate with GAP categories as expected. Mechanical failure rates did not progressively increase with higher GAP scores, and patients categorized as “proportionate” were not immune to complications. These findings have been attributed to population differences, as Asian patients generally have lower PI and reduced bone mineral density, suggesting that the original GAP thresholds may not be directly applicable. Subsequently, ethnicity-adjusted GAP criteria have been explored. One recent study proposed a modified GAP scoring system for a Chinese patient cohort, adjusting alignment targets to better reflect normative spinopelvic profiles [41].

Another limitation of the GAP system is its omission of key patient factors, such as age and bone quality, which are known to influence the risk of PJK/PJF. The original GAP score focuses primarily on radiographic alignment; however, advanced age or severe osteoporosis may predispose patients to junctional failure even when alignment is ideal. Therefore, some authors have attempted to improve the predictive performance of the original GAP scoring system by incorporating variables such as bone density or body mass index [42,43]. This limitation has been also been highlighted by multiple authors. A systematic review reported that across 15 studies (over 2,000 patients), higher mean GAP scores were associated with mechanical complications by only a small margin (mean difference=0.57 points between patients with and without complications) [44]. Specific components of the GAP score have also been questioned. The LDI was originally considered optimal when ranging between 50%–80%. However, recent data suggest that even within this “normal” LDI range, patients can develop PJF. Tobert et al. [45] reported no significant differences in junctional failure rates among patients classified as hypolordotic, aligned, or hyperlordotic based on LDI, calling into question the utility of LDI for predicting PJF. In a risk-factor analysis of 196 patients who all achieved age-adjusted alignment targets, Park et al. [46] found that an excessively high LDI (reflecting disproportionate concentration of lordosis in the lower lumbar spine) was a strong independent risk factor for PJF. Notably, the mean LDI (72%) among patients who developed PJF in that study remained within the “ideal” 50%–80% range defined by the original GAP system. Key results are summarized in Table 1.

In 2005, Roussouly et al. [47] described a qualitative classification of sagittal spine morphology. The Roussouly classification divides normal adult spines into four types (Types 1–4) based on sacral slope (SS) and the inflection point of lordosis [47]. In the setting of deformity correction, this classification is used to guide both the amount and distribution of lordosis restoration. The underlying hypothesis is that a mismatch between a patient’s inherent sagittal profile and the surgical correction may concentrate stress at junctional levels, thereby increasing the risk of failure. Therefore, restoring the spine to a patient’s ideal Roussouly profile may help minimize junctional stress and reduce the risk of PJK/PJF. Because the original Roussouly classification was derived from spinal morphology (i.e., SS and inflection point of lordosis) in asymptomatic healthy individuals, applying it directly to patients with established spinal deformities presents clinical challenges. Therefore, several researchers have proposed determined a patient’s Roussouly type based on PI values (i.e., theoretical types; categorized as PI <45°, 45°–60°, and >60°) given that PI remains constant even in the presence of deformity (Fig. 4) [48–50]. Postoperative Roussouly types (i.e., current types; categorized as SS <35°, 35°–45°, and >45°) are then determined using SS and lumbar apex. Comparison of preoperative and postoperative current types allows assessment of whether sagittal correction aligns with the Roussouly framework.

Multiple retrospective studies, as well as a recent meta-analysis, suggest that Roussouly-aligned correction is associated with significantly lower rates of PJK, PJF, and related revision surgery following ASD correction. A multicenter study of 290 patients with ASD reported that those whose postoperative alignment matched ideal Roussouly profile (based on their PI) had a mechanical complications rate of 22.5%, compared with 46.8% among patients who were not appropriately aligned [48]. Failure to achieve Roussouly-based alignment targets was associated with a risk of mechanical failure (relative risk of 3.0), with PJK being the most common complication (18% incidence). Similar results have been reported in subsequent studies [49,51–53]. A systematic review and meta-analysis of 10 studies (1,454 patients) demonstrating that patients whose postoperative alignment matched to their ideal Roussouly type had fivefold fewer mechanical complications than unmatched patients (OR, 0.22; 95% confidence interval [CI], 0.12–0.41) [52]. In particular, unmatched patients had higher odds of developing PJK (OR, 1.6) and rod breakages (OR, 1.75). Interestingly, despite the reduction in radiographic complications, no significant difference in overall reoperation rates was observed between matched and unmatched groups (OR, 0.48; p=0.14). Although several studies support the predictive value of the Roussouly classification, some limitations have been reported.

Regarding HRQOL, Pizones et al. [49] reported that matched correction based on Roussouly type was associated with superior clinical outcomes, with significantly better 2-year ODI and SF-36 PCS scores compared with unmatched patients. However, Passias et al. [50] found no significant differences in clinical outcomes such as minimal clinically important difference in ODI, EQ5D, and VAS scores between matched and unmatched groups. However, they observed that patients who both matched Roussouly sagittal spinal type and improved their SRS–Schwab modifiers achieved superior patient-reported outcomes.

Despite these encouraging findings, other studies have questioned the robustness of the Roussouly classification as a standalone predictive of PJK/PJF [50,54]. Several limitations warrant consideration. First, the meta-analysis by Aoun et al. [52] found no significant difference in revision surgery rates between matched and mismatched patients, suggesting that although Roussouly-aligned correction may reduce radiographic PJK, it may not fully prevent severe junctional failures requiring revision. This may reflect the influence of additional factors such as osteoporosis, fusion length, upper instrumented vertebra (UIV) level, or preventive techniques, which may contribute to PJF even when sagittal alignment is optimal. Second, although theoretical Roussouly types can be estimated based on PI categories (Fig. 4), distinguishing between Types 1 and 2 preoperatively may be challenging in patients with low PI (<45°) and established deformity, as deformity progression may not follow normal rules. Third, while the Roussouly classification describes the shape and distribution of lumbar lordosis, it does not directly quantify lordosis magnitude, such as LL or PI–LL, parameters that are strongly associated with PJK/PJF risk. Overall, the Roussouly classification provides patient-specific alignment guidance and is associated with lower PJK/PJF rates, but its limitations are evident in cases where other factors dominate risk or where a static classification cannot capture alignment nuances. The Roussouly concept should therefore be applied as part of a multifactorial risk mitigation strategy rather than used as a sole predictive tool. Key results are summarized in Table 1.

Although SVA reflects global sagittal balance, it is a linear parameter that requires calibrated radiographs and is sensitive to patient positioning or pelvic compensation. VPA refers to the angle formed between a line from the center of the femoral heads (hip axis) to a given vertebral body and a line from the hip to the S1 endplate [55]. T1PA was introduced by Protopsaltis et al. [56] to better capture global alignment while minimizing the influence of pelvic compensation and patient positioning (Fig. 5). This single metric incorporates both the patient’s forward inclination (i.e., T1 tilt that is angular component of SVA) and PT in one measurement [57]. Protopsaltis et al. [58] demonstrated that T1PA correlates strongly with HRQOL and proposed it as a useful surgical planning target. In a cohort of 559 patients with ASD, T1PA showed a linear relationship with disability. The authors recommended a target T1PA <14° for surgical correction to achieve minimal disability (ODI <20). Because T1PA is an angular measure, it is not affected by X-ray magnification and remains stable despite changes in compensatory mechanisms. As a result, intraoperative measurements can reliably reproduce postoperative values. However, this advantage is most applicable when fusion constructs extend to at least the upper thoracic spine, as stopping at the lower thoracic levels may allow postoperative changes in unfused thoracic segments. The T1PA concept has been extended to other vertebral levels to better describe overall spinal shape. Angles such as T4PA, T10PA, and L1PA have been defined to characterize the distribution of thoracic kyphosis and lumbar lordosis [55,58–62]. These segmental VPAs demonstrate excellent reproducibility, with recent reliability studies reporting intra- and inter-rater interclass coefficients greater than 0.90 for all measurements [55]. While these granular measurements are primarily used for research purposes, a practical clinical application has been the prediction of PJK using the upper instrumented vertebra-PT angle (UIVPTA). UIVPTA measures the pelvic angle to the uppermost instrumented vertebra of a fusion construct [62]. It effectively gauges how far forward (or backward) UIV is relative to the pelvis. Lower UIVPTA values indicate a more posteriorly inclined UIV, which has been associated with an increased risk of PJK. Proposed cutoff values for UIVPTA are 4.0° for patients with PI <45°, 9.5° for those with PI between 45° and 60°, and 13.0° for patients with PI greater than 60°. Similarly, the T10-pelvic angle (T10PA) has been proposed by the International Spine Study Group as a fixed angle within the fusion (Fig. 5) [60]. Normal T10PA values were found to correlate with PI (r=0.533) and age (r=0.308), leading to the proposal of PI- and age-adjusted T10PA targets with acceptable offsets of ±3.5°. Patients corrected within this functional range demonstrated improved clinical outcomes and lowest rates of PJK. Once the UIV extends to T10 or above, the T10PA is not affected by the occurrence of PJK. Surgeons can optimize T10PA through appropriate contouring of lumbar lordosis and thoracic kyphosis to reduce junctional stress. Preserving adequate upper spinal curvature (instead of making the upper spine too straight) may increase T10PA and thereby possibly reduce the risk of PJK [63–65].

Hills et al. [61] described the T4–L1–hip axis paradigm by studying normal adult volunteers without spinal pathologies to define regional and global sagittal alignment relative to PI using parameters that remain fixed within an instrumented fusion construct (Fig. 6). The T4–L1–hip axis consists of two components: a standard L1PA and the T4–L1PA mismatch. The authors identified a relatively strong relationship between L1PA and PI (r²=0.58), as well as between T4PA and L1PA (r²=0.81). According to their results, approximately 80% of individuals demonstrated an L1PA=0.5×PI−21°, with T4PA within 4° of L1PA. When T4PA equals L1PA, a line drawn from the center of T4 to the hip axis intersects the center of L1 (T4–L1PA mismatch=0°). This paradigm was subsequently validated with respect to mechanical complications in a cohort of 247 adults who underwent extensive spinal fusion from the upper thoracic spine (≥T4) to the sacrum [66]. After controlling for age, the authors modeled deviations in L1PA and T4–L1PA mismatch and found that deviation from a standard L1PA or T4–L1PA discrepancy in either direction, overcorrection or undercorrection, was correlated with an elevated risk of mechanical complication. Mechanical failure risk was minimized when L1PA was within ±2° of PI×0.5−19° (i.e., within 0° to 4° of the normative L1PA) and when the T4–L1PA mismatch ranged between −3° and +1°. Key advantages of these parameters include their continuous nature rather than categorical classification and the fact that both T4PA and L1PA can be measured intraoperatively and remain stable following long fusion constructs extending to the sacrum. Nonetheless, further external validation is needed to enhance comprehension. Key results are summarized in Table 1.

This study is a narrative review rather than a systematic review; therefore, not all publications related to sagittal alignment were included. The literature synthesis primarily focused on influential studies published in major spine journals, and relevant studies outside these sources may not have been captured. This selective approach may limit the overall comprehensiveness of the review. In addition, with the exception of the SRS–Schwab classification, most alignment schemes were originally developed with an emphasis on predicting mechanical complications rather than evaluating clinical outcomes. Consequently, comparative evidence regarding HRQOL outcomes across different alignment strategies remains less robust than evidence related to mechanical complications. Finally, PJK/PJF are multifactorial conditions influenced by factors such as fusion length, prophylactic methods (e.g., hook, tether, and cement augmentation), soft-tissue dissection techniques, and bone quality. Because this article focused specifically on alignment paradigms, these non-alignment factors were not addressed in detail and should be considered when interpreting the conclusions.

Over the past two decades, tremendous progress has been made in understanding how sagittal alignment in ASD surgery influences both clinical outcomes and mechanical complications. Early observations linking positive sagittal balance to pain and disability have evolved into increasingly sophisticated classification and surgical planning systems. The SRS–Schwab classification established clear alignment thresholds associated with HRQOL, highlighting the importance of parameters such as PI–LL, PT, and SVA. Building on this foundation, age-adjusted alignment goals recognize that optimal alignment is not universal, demonstrating that a less aggressive correction in older patients can preserve quality of life improvements while reducing the risk of PJK. The GAP score introduced a proportionality paradigm, showing the need to harmonize alignment with pelvic anatomy to mitigate mechanical failure. The Roussouly classification further underscores the importance of respecting each patient’s inherent sagittal morphology, as appropriate lordosis distribution for a given PI is associated with more stable and durable reconstructions. Finally, VPA, including T1PA and the T4–L1–hip axis, provide practical angular targets that are independent of patient positioning, can be measured intraoperatively, and have demonstrated associations with HRQOL and PJK risk. Given that no single alignment is universally optimal, a thorough understanding of the strengths and limitations of each framework is essential for individualized surgical planning. Future advances in imaging, patient-specific modeling, and artificial intelligence may further refine the prediction of optimal sagittal alignment and junctional complication risk. In addition, prospective multicenter studies are needed to validate existing frameworks and to develop comprehensive alignment strategies that integrate biological and functional factors alongside radiographic parameters.