Trends in the surgical treatment for metastatic spinal tumor in Japanese administrative data between 2012 and 2020.

Background. Both cancer diagnosis/treatment modality and surgical technique for the spine have been developed recently. Nationwide trends in the surgical treatment for metastatic spinal tumors have not been reported in the last decades. This study aimed to examine recent trends in the surgical treatment for spinal metastasis and in-hospital patient outcomes using nationwide administrative hospital discharge data. Methods. The Diagnosis Procedure Combination database from 2012 to 2020 was used to extract data from patients who underwent surgical procedures for spinal metastasis with the number of non-metastatic spinal surgery at the institutions that have performed metastatic spine surgeries at least one case in the same year. Trends in the surgical treatment for spinal metastasis, patients’ demographics, and in-hospital mortality/outcomes were investigated. Results. This study analyzed 10321 eligible patients with spinal metastasis. The surgical treatment for spinal metastasis increased 1.68 times from 2012 to 2020, especially in fusion surgery, whereas the proportion of metastatic spinal surgery retained with a slight increase in the 2%s. Distributions of the primary site did not change whereas age was getting older. In-hospital mortality and length of stay decreased over time (9.9–6.8%, p < 0.001; 37 days–30 days, p < 0.001). The postoperative complication and unfavorable ambulatory retained stable and slightly decreased, respectively.


Introduction
The overall cancer incidence in the age-standardize rate increased until around 2010 and stabilized thereafter, whereas cancer mortality has been decreasing in Japan as well as the United States.[1,2] The development of systemic treatment has substantially improved the overall life expectancy of patients with cancers.The most frequent site of bone metastasis is the spine and approximately 50% of bone metastasis involves to spine.[3] Spinal metastasis occurs in 20-40% of patients with cancer and up to 20% of them become symptomatic.[4][5][6] The clinical presentation of spinal metastasis is predominantly axial pain, neurological de cit, and progressive deformity.The increased survival of patients with cancer corresponds to the increased symptomatic spinal metastasis.Treatment for spinal metastasis is multidisciplinary involving chemotherapy, corticosteroids, bone-modifying agent, radiotherapy, and surgery.[7] The majority of surgical treatment for spinal metastasis is palliative surgery, including spinal fusion procedure with or without decompression, decompression alone procedure, or percutaneous vertebroplasty, to maintain or improve the performance status (PS) of quality of life (QOL) by providing pain relief or maintenance/recovery of neurological functions.[8,9] Total en bloc spondylectomy (TES) is indicated for limited patients with expected longer survival and controlled primary tumor without extraspinal metastasis.[10] Indications for spinal surgery for degenerative diseases have widened considerably in recent decades with technological advances and improved surgical techniques, therefore, the number of spinal surgery have increased by year in Asian countries as well as United States or European countries.[11][12][13][14] Surgical treatment for spinal metastasis has also developed in the last decade, by developing surgical techniques of minimally invasive surgeries [15] or increased medical evidence from clinical trials [16], as well as developing treatment of cancer.Despite the recent increasing number of cancer patients and developing treatment of spinal metastasis, recent nationwide trends which include surgical volume/procedure, patients' background, and outcomes have not been reported and are needed for informed surgical decision-making and healthcare management.
Accordingly, the present study aimed to examine trends in the surgical treatment for spinal metastasis and in-hospital patient outcomes in Japan from 2012 to 2020, by analyzing nationwide administrative hospital discharge data collected for the Diagnosis Procedure Combination (DPC).

Data Source and Patient Selection
The Diagnosis Procedure Combination (DPC) database is a national inpatient database in Japan which is organized by the DPC Study Group.This database includes data on approximately 7 million inpatients from > 1000 hospitals annually and covers approximately 50% of acute-care inpatients in Japan.The database includes age and sex; diagnoses, comorbidities at admission, and complications after admission recorded with Japanese text and the International Classi cation of Diseases, 10th Revision (ICD-10) codes; procedures; drugs and devices used; hospital stay duration; activities of daily living described using the Barthel Index (at both admission and discharge); and admission and discharge status.
The institutional review board of our institute approved our study design.Informed consent from individual patients was waived because of the anonymous nature of the data.
This study used inpatient data of patients discharged from April 1, 2012, to March 31, 2021.Data of patients who underwent spinal surgeries for spinal pathological spine fractures or bone metastasis (ICD-10 code, M8448 or C795) were included.Patients who underwent multiple spine-related surgeries, including two-step surgery, wound infection surgery, or implant failure surgery were included because the database could not distinguish between planned staged surgery and unplanned incidental surgery.We excluded patients with the following: (1) the main pathology for the spinal procedure was another spinal non-tumor disease (degenerative disorders, spinal deformity, in ammatory disorders, or spinal trauma) from diagnosed codes; (2) the procedure code was the spinal/pelvic tumor resection procedure because the code usually has been used in the case of benign tumors; (3) diagnosis recorded with ≥ 2 primary cancer because the primary tumor that caused the bone metastasis could not be identi ed.
We also exacted data regarding whole spinal surgery regardless of spinal metastasis at the institutions that have performed metastatic spine surgery in at least one case in the same year to assess the association between the number of whole spinal surgery cases and metastatic spine surgery cases.
The following data were extracted from each patient with metastatic spinal surgery: treatment institution according to whole spinal surgery volume, site of primary cancer, age, gender, body mass index, surgical methods, comorbidities at admission, non-skeletal metastases, Barthel Index at admission and discharge, in-hospital mortality, length of stay, postoperative complications, and destination after discharge.
The spinal surgery volume of the institution was divided into low (< 100 cases/year), middle (100-300 cases/year), and high-volume (≥ 300 cases/year) considering the annual number of the cases of the median and interquartile range (IQR).
We divided the surgical method into four categories as follows: Total en bloc spondylectomy (TES), fusion surgery (palliative instrumentation surgery with or without decompression, including anterior and posterior approach), laminectomy alone procedure (including laminoplasty), and vertebroplasty (including kyphoplasty).Surgical methods were categorized based on the type of initial surgery regardless they received more than one surgery.
The site of the primary tumor was extracted according to the ICD-10 code as follows: lung, prostate, breast, kidney, colon, and others.An unspeci ed primary site was de ned as a primary site that could not be identi ed from DPC data in this study, which might include "unknown primary cancer."We extracted the following metastases: lung (C780-C782), liver (C787), brain (C793), and others (C770-C779, C784-C786, C788, C790-C792, and C796-C799).Non-skeletal metastasis was de ned as distant metastasis other than bone, including visceral, lymph nodes, brain, and disseminated metastases.We divided patients based on primary site into those with and without rapid tumor growth based on the New Katagiri score [17].The rapid growth group included patients with lung, gastric, esophageal, pancreatic, and urothelial carcinoma.
The comorbidities was evaluated by the Charlson Comorbidity Index (CCI) [18] which was calculated based on Quan's protocol [19] and di cult anesthesia.The list of the ICD-10 codes used to identify the comorbidities in CCI are shown in Supplement Table 1.This study used modi ed CCI which excluded the input for history of malignancy and metastatic disease according to the previous report [20].Anesthesia di culty was de ned by Japanese standardized reimbursement anesthesia codes of L0008-5 (Supplemental Table 2).a Sub-score for mobility in the Barthel Index at admission was 10 or 15.
c Visceral, lymph nodes, brain, and disseminated metastasis other than bone.
d Instrumentation with and without decompression via anterior or posterior approach.
e Including balloon kyphoplasty.

Outcomes
The outcomes of this analysis were evaluated by (1) in-hospital mortality with the 30-day mortality, (2) length of stay, (3) the proportion of discharge to home, (4) postoperative complication, and (5) unfavorable ambulatory status.Length of stay was calculated by the duration between the initial spinal surgery and the discharge.The postoperative complications included acute coronary syndrome, heart failure, respiratory disorder, sepsis, pulmonary embolism, stroke, renal failure, and urinary tract infection classi ed by using ICD-10 codes (Supplemental Table 1).
The outcomes of gait ability were evaluated by unexpected ambulatory status for surgical treatment for spinal stenosis."Unfavorable" was de ned as (1) non-ambulatory at discharge or (2) ambulatory status deterioration calculated according to a previous report [21].

Statistical analysis
The results are presented as median (IQR).Initially, tends of number of patients who underwent surgeries for metastatic spinal tumors and non-metastatic spinal tumor was investigated including the number of investigated institutions by year.Subsequently, trends of patients' demographics including the primary site of the tumor, surgical methods, and outcomes were investigated in patients who underwent spinal surgery for a metastatic spinal tumor.Jonckheere-Terpstra trend test and Cochran-Armitage test for trend were used to examine the annual trends in continuous and categorical variables, respectively.
Sensitivity analysis was performed for trends of metastatic and non-metastatic spinal surgery by analysis limited to the institution that performed metastatic spinal surgery every year.Trends of length of stay after spinal surgery for spinal metastasis were also performed a sensitivity analysis limited to patients who were discharged to home.
A p-value of < 0.05 was considered statistically signi cant.All statistical analyses were performed by SPSS Statistics for Windows, Version 26.0 (IBM Corp., Armonk, NY, USA).

Results
This study included 10,321 patients who underwent metastatic spinal surgery between 2012 and 2020 with 473,391 patients who underwent non-metastatic spinal surgery at institutions that performed metastatic spinal surgery in the same year, as shown by the ow diagram (Fig. 1).Patients' demographics through data of nine years were shown in Table 1: median age of 68 (60-74) years, male 6505 (63%).

Trends in metastatic spinal surgery and patients' demographics
The number of patients who underwent spinal surgery for spinal metastasis increased over time from 877 in 2012 to 1,479 in 2020 (Fig. 2A, supplemental Table 3p < 0.001).The metastatic spine surgery was performed in a total of 721 institutions during the study period.Meanwhile, the number of patients who underwent spinal surgery for non-metastatic disorders in these institutions increased over time (Fig. 2A, supplemental Table 3p < 0.001).The proportion of metastatic spinal surgery to whole spine surgery ranged from 2.0 to 2.6% during the study period (Fig. 2B).The number of treated hospitals for spinal metastatic surgery did not change during the study period, while the rate of high-volume spine surgery centers increased from 9% in 2012 to 16% in 2020 (Supplemental Table 3).In the sensitive analysis that was limited to the institution which performed metastatic spinal surgery every year, the number of patients who underwent surgery for metastatic spinal tumor and non-metastatic spinal tumor was also increased over time (Supplement Fig. 1).
Trends in surgical methods are shown in Fig. 3. Fusion surgeries which occupied the most of surgical methods of > 80% of metastatic spinal surgery, increased over time (84.5% in 2012, and 87.2% in 2020, p = 0.01).The ratio of TES decreased over time from 2.6% in 2012 to 1.0% in 2020 (p < 0.001).The decompression alone procedure also decreased over time form 7.4% in 2012 to 5.8% in 2020 (p < 0.001).
Figure 4. shows trends in patients' demographics and Table 2 demonstrates trends in the primary site of the tumor.Age at the surgery was getting older each year (≥ 70 years: from 36.3% in 2012 to 48.7% in 2020, p < 0.001).The distribution of gender did not change (p = 0.253).The ratio of patients with modi ed CCI ≥ 1 decreased from 37.9% in 2012 to 31.2% in 2020, p < 0.001), whereas the ratio of anesthesia di culty has not changed through the period (p = 0.29).The distribution of the primary site of the tumor has not changed each year: lung, prostate, breast, and kidney, were the most common primary sites for spinal metastasis (Table 2).

Trends in in-hospital outcomes after metastatic spinal surgery
The overall in-hospital outcomes through the periods are shown in Fig. 5 and Table 3. In-hospital mortality decreased over time from 9.9% in 2012 to 6.8% in 2020 (p < 0.001), and the 30-day mortality also decreased over time from 7.1% in 2012 and 4.5% in 2020 (p < 0.001).The length of stay shortened over time from 37 (22-56.5)days in 2012 to 30 (18-45) days in 2020 (p = 0.008).The length of stay limited to patients who discharged to home was also indicated shortened trends over time (Supplemental Fig. 2).There was no signi cant change in the proportion of discharge to home and overall postoperative complication.The unfavorable ambulatory at discharge decreased slightly but signi cantly from 39.0% in 2012 to 36.9% in 2020 (p = 0.03).

Discussion
This study demonstrates increasing trends in surgical treatment for spinal metastasis, as well as increasing trends in surgical treatment for other spinal disorders in Japan during the last decade.The number of patients who underwent metastatic spinal surgery increased 1.68 times from 2012 to 2020, whereas the proportion of metastatic spine surgery in overall spinal surgery mostly retained with a slight increase in the 2%s.
The treatment paradigm may have been shifted to surgical treatment inspired by the landmark 2005 randomized controlled trial by Patchell et al.[8] which demonstrated improved ambulatory status and preserved neurological function in patients who had undergone surgery followed by radiation as compared to radiation alone.[22][23][24] Current advance of cancer diagnosis/treatment modality enables early diagnosis and early treatment and lead to decrease mortality by cancer in the all over the world.[1,2] Therefore, the metastatic spine surgery had not increase according to the speci c primary site of tumor despite of increase of the number of spinal surgery.[25,26] Contrary to their reports which investigated between 2005 and 2014 in US, this study demonstrates an increase of metastatic spinal surgery over time.It may be explained that recent advances in spinal surgery especially in minimally invasive surgery [15] reduce the risk involved in the surgical treatment and lead changes in practice patterns and patient selection criteria for surgical treatment, although more detailed investigations are needed.
This study demonstrated a change in the proportion of surgical procedures.The proportion of fusion surgery increased and decompression alone surgery decreased over time, similar to the previous trends in the patients with lung or breast cancer metastasis [25,26].A possible reason for this trends is the establishment of spinal instability neoplastic score(SINS) [27].SINS were advocated in 2010 by the Spine Oncology Study Group to assess the degree of spinal instability caused by metastatic disease.The score consists of the sum of 5 radiographical and 1 clinical item, resulting in a total score between 0 and 18 points.[27] Although the optimal cutoff value for the need for fusion surgery has not been reached consensus, it is generally divided into 3 categories: stable (0-6 points), impending/potentially unstable (7-12 points), and unstable (13-18 points).With the worldwide recognition of SINS, spine surgeons or oncologists become to determine the application of fusion procedures by the score of SINS [28,29].
Combined with the advance of minimally invasive fusion procedures, the spread of SINS may have led to an increase in spinal fusion procedures, which were formerly considered highly invasive for metastatic patients and often avoided.
We hypothesized that the distribution of the primary site of tumors might change through the period because of drastic changes in chemotherapy for each cancer.However, it did not uctuate in this study period from data of DPC database.Therefore, further detailed clinical research including the subtype of cancer of the primary site are needed to elucidate the in uence of recent advance in cancer diagnosis/treatment modality.
Among trends in patients' demographic, this study showed that age at the surgery was getting older in consent with previous reports according to advanced surgical procedure or increased medical evidences.[23,24,30] However, preoperative comorbidity evaluated anesthesia di culty and modi ed CCI in this study did not increase over time, contrary to previous reports which demonstrate increased patients with comorbidity by year [23,24].It could be several explanations why patients with comorbidity did not change over time.First, the previous studies [23,24] used different evaluation for comorbidity of Elixhauser comorbidity index.The Elixhauser comorbidity index which includes a set of 30 medical comorbidities have been reported to outperform the CCI to predict mortality or complications.[31] Second, both comorbidity criteria used in this study: anesthesia di culty and modi ed CCI ≥ 1, might be too severe to evaluate patient selection criteria for spinal surgery by year.Therefore, more detailed information on comorbidity is needed to clarify trends in patients' comorbidity in surgical treatment.
In-hospital outcomes showed several trends from 2012 to 2020.In-hospital mortality decreased 3.1% from 9.9% in 2012 to 6.8% in 2020.The length of stay was shortened median of 7 days from 37 days in 2012 to 30 days in 2020.Meanwhile, the proportion of discharged to home, and postoperative complications retained during the last decade.The postoperative unfavorable ambulatory slightly decreased.Because in-hospital mortality is affected by the length of stay, we analyzed the 30-day mortality, which resulted in a signi cant decrease over time.Taken into consideration together, shortterm mortality after surgical treatment for spinal metastasis decreased over time and postoperative outcomes for ambulatory improved, despite of increase in surgical cases.This might arise from the recent development of the multidisciplinary approach for spinal metastasis [32,33], as well as the medical advance of surgical technique and postoperative care.
This study has several limitations.First, DPC diagnoses may be less validated than those in prospective cohorts.Although the accuracy of diagnosis has been relatively high in terms of metastatic cancer, and the diagnostic records in the DPC database had a sensitivity of 73.2% and a speci city of 96.0% for detecting metastatic tumors, comorbidity diagnosis coding showed lower sensitivity than primary diagnosis [34].However, reporting might not vary substantially within the database.Second, participating institutions in the DPC database at the year might affect the analysis for trends by year.However, this response bias should be small in this study, because the number of participating institutions did not vary in the study periods as shown in Supplement Table 3. Lastly, the DPC data were limited to in-hospital events; consequently, the true incidence rates of complications and mortality may have been underestimated.Despite these limitations, we believe that these data provide a reasonable picture of trends and in-hospital patient outcomes in the surgical treatment for spinal metastasis in Japan during the last decade.
In conclusion, surgical treatment for spinal metastasis increased, especially in fusion surgery, in Japan during the last decade.Age at the surgery was getting older, whereas the distribution of the primary site of tumors and preoperative comorbidity did not change.In-hospital mortality and length of stay decreased over time, whereas discharge to home and postoperative complications remained stable.The postoperative unfavorable ambulatory decreased slightly.Although further investigation is necessary to validate the long-term e cacy of surgical treatment for spinal metastasis in increasing trends by assessing the patient's quality of life after surgery, the results of this study may provide useful information for healthcare policy or clinical research planning.

Figure 1 Flow diagram of the study Figure 2
Figure 1

Figure 3 Distribution
Figure 3

Figure 4 Patients
Figure 4

Table 1
Demographics of patients who underwent metastatic spinal surgery between and 2020.

Table 2
Trend of distribution of primary site from 2012 to 2020

Table 3
Characteristics of in-hospital outcomes after metastatic spinal tumor between 2012 and 2020 a Non-ambulatory at discharge or deterioration of sub-score for mobility in the Barthel Index.