Long Term Clinical Outcomes of Patients with Colorectal Cancer with Metastatic Epidural Spinal Cord Compression treated with Hybrid Therapy (Surgery followed by Stereotactic Body Radiation Therapy)

Abstract

Background

“Hybrid-therapy”, consisting of separation-surgery followed by stereotactic body radiation therapy (SBRT) has become the mainstay treatment for radioresistant spinal metastases. Histology-specific outcomes for hybrid therapy are scarce. In clinical practice, colorectal cancer (CRC) is particularly thought to have poor outcomes regarding spinal metastases. The goal of this study is to evaluate clinical outcomes for patients treated with hybrid therapy for spinal metastases from CRC.

Methods

This is a retrospective study performed at a tertiary cancer center. Adult patients with CRC spinal metastasis who were treated with hybrid-therapy for high-grade epidural spinal cord or nerve root compression from 2005-2020 were included. Outcome variables evaluated included patient demographics, overall survival (OS) and progression-free survival (PFS), surgical and radiation complications, and clinical-genomic correlations.

Results

Fifty patients met inclusion criteria. Progression of disease occurred in 7 (14%) patients at the index-level requiring reoperation and/or reirradiation at a mean of 400 days after surgery. Postoperative complications occurred in 16% of patients, with 3 (6%) requiring intervention. Adenomatous polyposis coli (APC) exon 14 and 16 mutations were found in 15 of 17 patients tested and in all 3 of 7 local failures tested. Twenty patients (40%) underwent further radiation due to disease progression at another spinal levels.

Conclusions

Hybrid-therapy in CRC patients resulted in 86.7% local control at two years after surgery, with limited complications. APC mutations are commonly present in CRC patients with spine metastases and may suggest worse prognosis. Patients with CRC spinal metastases commonly progress outside the index treatment level.

Introduction

Colorectal cancer (CRC) is one of the most common cancers, with more than 1,000,000 new cases per year worldwide ¹. While the introduction of new treatments with targeted agents and immunotherapies has improved prognosis for multiple cancer types, the prognosis for metastatic CRC remains poor². Half of all CRC patients die from metastatic disease, and the overall 5-year survival for patients with Stage IV disease is 10.5% ³. Approximately 27% of CRC patients develop bone metastases, with the spinal column being the most common site ^4,5. The main sites of spinal metastases for CRC in descending order are lumbar (36%-75%), thoracic (17%-61%), sacrum (6%-35%), and cervical spine (2%-7%) ^6–8. With advances in cancer care, the median survival of CRC patients is expected to increase, and patients are more likely to experience vertebral column metastases leading to metastatic epidural spinal cord compression (MESCC) ⁹. Patients with MESCC can present with progressive neurological deficits, mechanical instability, and pain, all of which have the potential to severely affect a patients quality of life and their ability to perform activities of daily living ¹⁰.

Treatment of MESCC typically requires a multimodal approach including surgery and radiation therapy. Historically, conventional external beam radiation therapy (cEBRT) for CRC spinal metastasis resulted in a poor median survival (4.1 months) and local control ¹¹. Given the relative radioresistance of CRC spine metastases, recent treatment paradigms have utilized stereotactic body radiotherapy (SBRT) to provide highly conformal intensity-modulated radiation with an ablative dose ¹². Surgical indications were guided using the NOMS decision making framework in a multi-disciplinary approach.

Hybrid therapy, consisting of separation surgery (ie. posterolateral tumor decompression separating the tumor from the spinal cord) followed by SBRT has become the mainstay of treatment for radioresistant spinal metastases with high-grade MESCC ^13,14. The goals of treatment are to decompress the spinal cord, restore or preserve neurologic function, stabilize the spinal column, and create separation between the tumor and the spinal cord thus producing a safe target for the delivery of SBRT ^15,16. The traditionally poor prognosis associated with CRC causing MESCC remains challenging for spine surgical oncologists trying to balance the extent of surgical resection with the palliative goals of treatment. Herein we present clinical outcomes from hybrid therapy (i.e., separation surgery followed by SBRT) for CRC patients presenting with MESCC.

Materials and Methods

Study Design

This is a single-center, retrospective study performed at a tertiary cancer center. Adult patients (>18 years old) with pathologically confirmed CRC spinal metastasis who were treated with hybrid therapy from 2005-2020 were included. Patients were excluded if they received postoperative conventional radiation therapy or did not receive any post-operative radiation. Patients who were previously treated surgically at the index level were also excluded. Patients who received prior cEBRT or SBRT were included and were considered “salvage” hybrid therapy. The primary endpoint of this study was progression of disease from the date of surgery, defined as requiring additional surgery or radiation therapy at the index-tumor level. While other courses of care including hospice, end of life, or decision not to treat were considered, this cohort consisted of patients with a primary endpoint that was indicated for intervention. Follow-up consisted of routine MRI at 3-month intervals post-hybrid therapy and as clinical presentation would necessitate. This study was approved by the local institutional review board.

Outcome Variables

Preoperative variables studied included patient demographic data, prior treatment, neurological status through ASIA scores, and performance status via ECOG scores. Tumor variables included histology, tumor location and tumor mutational status when available. Spinal Instability Neoplastic Score (SINS) was used to determine mechanical instability, and the degree of MESCC was calculated using the epidural spinal cord compression scale.

Surgical Technique

All patients underwent separation surgery via transpedicular posterolateral decompression and posterior instrumented stabilization with pedicle or lateral mass screw-rod fixation and when indicated, cement augmentation of the pedicle screws. Perioperative variables such as open versus minimally invasive surgery, number of levels decompressed, number of levels instrumented, the use of cement augmented screws, cement vertebral reconstruction at the index level, and concomitant intraoperative use of a P32 brachytherapy plaque were recorded. All patients underwent a postoperative CT myelogram which was used for planning SBRT according to institutional protocols. The time to postoperative radiation, total treatment dose, and number of fractions were also recorded.

Following separation surgery, SBRT was administered to all included patients. SBRT contours were planned according to standard guidelines¹⁷. Briefly, the preoperative MRI was used to delineate the gross tumor volume (GTV) which included the intraosseous, epidural, and paraspinal components. The GTV coverage was contoured to the preoperative tumor volume for radiation treatment to define the dural margin, allowing a clear delineation of cerebrospinal fluid space and spinal cord in the presence of spinal instrumentation. The clinical tumor volume (CTV) was an expansion of the GTV contour to account for microscopic tumor via adjacent marrow spaces. The planning target volume (PTV) represented a minimal expansion of the CTV that accounts for uncertainties in radiation set up and delivery ¹⁸.

Statistical Analysis

Descriptive statistics including means, standard deviations, and proportions were used to characterize the cohort. For PFS, events were progression and death. PFS follow-up time was calculated in two separate ways to fully explore time from treatment: 1) from date of surgery until date of progression, death, or last follow-up, whichever occurred first and 2) from date of postoperative radiation until date of progression, death, or last follow-up, whichever occurred first. OS follow-up time was calculated similarly: 1) from date of surgery until death (events) or last follow-up (censored) and 2) from date of postoperative radiation until death or last follow-up. Cox proportional hazards regression modeling was performed to associate variables with OS and PFS from both date of surgery and date of postoperative radiation. Variables of interest associated with OS and PFS included age, LOS, EBL, construct length, SINS score, levels decompressed, sex, ECOG, index level of metastasis, ASIA, ESCC, and mutational status for genes, which were tested in enough patients and where the distribution of negative/positive mutational status allowed for modeling. All variables were modeled in a univariate fashion. Any variables meeting statistical significance in the univariate setting were entered into a multivariable model. Cumulative incidence of progression in the competing risk setting with death as the competing risk was calculated with corresponding 95% confidence intervals. All statistical tests were two-sided with an alpha level of significance <0.05. Analyses were performed in SAS v9.4 (The SAS Institute, Cary, NC) and R v4.0.4 (The R Foundation for Statistical Computing).

Results

Demographic Data

A total of 50 patients met inclusion criteria and were included in the final analysis. Median age at surgery was 55.2 years, and 32 (64%) were male. Thirty-five patients with high grade ESCC (ten (20%)) grade 2, and twenty-five ((50%) grade 3) and 13 patients with moderate grade ESCC (1c) required cervical (n=3) or lumbar (N=10) nerve root decompression.

Preoperative ECOG scores were 1 in 25 (50%), 2 in 16 (32%), 3 in 8 (16%), and 4 in 1 (2%). Preoperative ASIA scores were C in 1 (2%), D in 13 (26%), and E in 36 (72%). The median SINS score was 10 (range 6-14). Tumor location was thoracic in 21 (42%), lumbar in 15 (30%), and cervical spine in 5 (10%). Tumors were also located across junctional sites: thoracolumbar 5 (10%), cervicothoracic 3 (6%), and lumbosacral 1 (2%). (Table 1)

Table 1:

Preoperative Demographic and Neurologic Function

Variable	Characteristic	N
Sex	Male	32 (64%)
	Female	18 (36%)
Preop ECOG	1	25 (50%)
	2	16 (32%)
	3	8 (16%)
	4	1 (2%)
Preop ASIA	C	1 (2%)
	D	13 (26%)
	E	36 (72%)
ESCC Score	1C	7 (14%)
	2	14 (28%)
	3	27 (54%)
Index level	Cervical	5 (10%)
	Cervicothoracic	3 (6%)
	Thoracic	21 (42%)
	Thoracolumbar	5 (10%)
	Lumbar	15 (30%)
	Lumbosacral	1 (2%)
SINS Score	1-6	1 (2%)
	7-12	42 (84%)
	13-18	5 (10%)

Prior Treatment

One (2%) of the 50 patients had prior spine surgery for decompression in the setting of metastatic epidural spinal cord compression outside the treatment area and none had prior kyphoplasty at the index tumor level. Twenty-two (44%) patients underwent previous radiation at the index tumor level with varying doses and are considered “salvage hybrid therapy”.

Operative details (Table 2a)

Table 2a and 2b:

Operative Details. Surgical and Radiotherapy Outcomes

Surgical Outcomes	Median	Mean	St. Dev.	Min	Max
Length of Surgery, minutes	153	168.3	65.7	83	385
Construct Length	53	5.4	2	3	10
# Levels Decompressed	2.5	2.3	1.1	1	6
EBL	500	583	422.5	50	1800
a
RT Outcomes	Median	Mean	St. Dev.	Min	Max
Time to Post-op RT, days	21	25.1	11.8	10	55
Total Dose, Post-op RT (cGy)	2700	2859	409.5	2000	4000
Total Fractions, Post-op RT	3	3.5	1.2	1	5
Biologic Equivalent Dose (BED)	51.3	54.5	10.6	37.5	81.6
b

Forty-nine (98%) patients underwent traditional open separation surgery, while 1 (2%) patient underwent minimally invasive surgery. The median length of surgery was 153 minutes. The median construct length was 5 levels and the median number of levels decompressed was 2.5 levels. The median estimated blood loss was 500 mL. The majority (94%) of patients were discharged home, while 3 (6%) patients required acute inpatient rehabilitation. Poly-methyl-methacrylate (PMMA) bone cement augmented screws were used in 9 (18%) patients and intraoperative cement reconstruction of the vertebral body (ie vertebroplasty) was performed in 4 (8%) patients. Intra-operative brachytherapy with a P32 plaque was used in 5 (10%) patients.

Radiation Details (Table 2b)

All 50 patients underwent postoperative SBRT at the index tumor level. Twenty (40%) patients underwent radiotherapy at an adjacent level as well and 1 underwent additional radiotherapy at a distant spinal site concomitantly. The median time to postoperative radiation was 21 days. The median total dose was 27Gy over a median of 3 fractions.

Mutational Data (Table 3)

Table 3:

Tumor Mutational Status Data

EGFR exon 18	Negative	33 (66%)
	Positive	1 (2%)
EGFR exon 19	Negative	33 (66%)
	Positive	1 (2%)
EGFR exon 20	Negative	33 (66%)
	Positive	1 (2%)
EGFR exon 21	Negative	33 (66%)
	Positive	1 (2%)
HER2	Negative	2 (4%)
FGFR2	Negative	19 (38%)
	Positive	2 (4%)
FGFR3	Negative	22 (44%)
KRAS	Negative	23 (46%)
	Positive	15 (30%)
TP53	Negative	7 (14%)
	Positive	15 (30%)
BRAF	Negative	31 (62%)
	Positive	1 (2%)
PIK3CA	Negative	32 (64%)
	Positive	2 (4%)
APC exon 14	Negative	17 (34%)
	Positive	1 (2%)
APC exon 16	Negative	2 (4%)
	Positive	15 (30%)
GNAS	Negative	20 (40%)
	Positive	2 (4%)
NRAS	Negative	32 (64%)
	Positive	2 (4%)

^*Complete genetic profiling was only acquired based on the treating medical oncologist’s request if there was a clinically significant impact on management of systemic therapy.

A diagnostic molecular pathology report evaluating mutational status was available in 38 (76%) patients. The full list of mutations that were tested included: EGFR exon 18, EGFR exon 19, EGFR exon 20, EGFR exon 21, HER2, FGFR2, FGFR3, BRAF and BRAF Exon 15, PIK3CA, APC exon 8, APC exon 14, APC exon 16, KRAS, GNAS, NRAS, BRCA2, TP53, and DNMT3A. APC exon 16 data was available in 17 patients of which 15 was positive (30%). KRAS mutation was present in 15 (30%) patients, negative in 23 (46%), and not recorded in 12 (24%) patients. TP53 mutation was present in 15 (30%) patients, negative in 7 (14%), and not recorded in 28 (56%) patients. Overall survival and PFS from date of surgery or from date of radiation was not significantly affected by tumor mutational status in this analysis.

Outcomes/Local tumor control

Progression of disease was documented in 7 (14%) patients at the index tumor level requiring reoperation and/or reirradiation at an average timepoint of 400 days after surgical intervention (range: 70-840 days). Two-year local control rate following surgery was 86.7% (95% CI: 76.4%-97.0%) (Figure 1). Twenty patients underwent further radiation due to disease progression at another spinal level. Postoperative complications occurred in 9 (18%) patients, 3 (6%) of which required surgical intervention. Revision surgery included 2 for wound revisions and 1 for hardware failure. Other complications included DVT, pneumonia, urosepsis and seroma (2) that were managed conservatively. Four patients (8%) suffered mild postoperative radiotherapy complications including fatigue, nausea, vomiting and transient worsening of leg pain. There were no incidents of radiation induced myelopathy in this cohort.

Figure 1:

Cumulative Incidence of tumor progression and death at time in months from radiotherapy

Predictors of clinical outcomes:

Preoperative ASIA D score was associated with a worse OS post-surgery (HR:2.07, 95%CI:1.02-4.19, p=0.04) and post-radiation (HR:2.18, 95%CI:1.07-4.43, p=0.03) when compared to Preoperative ASIA E. Preoperative ECOG of 2 trended towards worse OS post-surgery (HR:1.87, 95%CI:0.96-3.63, p=0.06) and post-radiation (HR:1.92, 95%CI:0.99-3.72, p=0.05) when compared to ECOG score of 1. Longer construct length was associated with worse PFS from date of surgery (HR:1.20, 95%CI:1.03-1.40, p=0.02) and from date of postoperative radiation (HR:1.20, 95%CI:1.03-1.41, p=0.02). Patients with lumbar disease had better PFS from date of surgery and post-radiation compared to those with cervical metastases (HR:0.33, 95%CI:0.11-0.96, p=0.04, and HR:0.32, 95%CI:0.11-0.95, p=0.04, respectively). Other variables not associated with PFS and OS in the univariable setting included age, length of surgery, estimated blood loss, preoperative SINS score, levels decompressed, sex, preoperative ESCC score, P32 brachytherapy use, and KRAS or TP53 mutations. In the multivariable setting for PFS from date of surgery, HR estimates were of similar magnitude and direction as the univariable setting for lumbar disease (HR:0.49, 95%CI:0.12-1.93, p=0.31) and length of construct (HR:1.12, 95%CI:0.90-1.40, p=0.32), though p-values were no longer statistically significant. Similar findings were observed for PFS post-radiation for lumbar disease (HR:0.47, 95%CI:0.12-1.82, p=0.28) and length of construct (HR:1.12, 95%CI: 0.90-1.40, p=0.31).

Discussion

This is the largest reported series of patients with MESCC secondary to CRC spine metastases that were treated with hybrid therapy. Surgery for metastatic disease is performed for palliative indications including pain relief, preservation of neurologic function, and improvement in quality of life. Each patient was discussed in a multidisciplinary tumor board including neurosurgery, radiation oncology and interventional radiology providers. Surgical decision making was performed utilizing the NOMS decision making framework for high grade spinal cord or nerve root compression ¹⁰. Surgical decision making was performed utilizing the NOMS framework (with ASIA and ESCC scores representing neurological considerations (N), expected response to available therapies as the oncologic assessment (i.e. CRC is a tumor that is radioresistant to conventional RT and thus the need for post operative SBRT (O), spinal mechanical instability as represented with SINS score (M) and multidisciplinary discussions are carried out with medical and radiation oncology to coordinate surgical intervention and to minimize disruption of chemotherapy or other systemic treatments ensuring patients can tolerate the proposed therapy (S). To note, we currently do not consider mutational status for patients with CRC for surgical decision making at this time. Our current data demonstrates that this hybrid therapy strategy provides a well-tolerated, reproducible treatment that leads to durable local control of CRC metastatic patients.

Early work by Brown et al and Leach et al documenting their experience with management of spinal metastases of CRC origin were valuable to better understand the clinical course ^11,19. Brown, et. al. reported outcomes of external beam radiotherapy (EBRT) in 34 patients with metastatic CRC presenting with ESCC, demonstrating a median overall survival of 4.1 months. Factors associated with favorable outcomes and improved survival included delivery of greater than 30 Gy RT, primary rectal site in origin, and patients who were fully ambulatory at the time of diagnosis ¹¹. Leach, et al. described their surgical-only experience of spinal metastasis of CRC origin. In their series of 4 patients, 2 had no change in neurological exam, and 3 had an average expected survival of 1.3 months ¹⁹. These two case-series demonstrate the limited value stand-alone therapy has in the control rates and survival of patients with spine metastasis from CRC.

Given the limited up-to-date literature on outcomes of spinal metastatic disease in patients with CRC, recent case series on MESCC, regardless of pathology can be extrapolated to better understand the impact of hybrid therapy. Laufer, et. al. described their experience with 186 patients treated with separation surgery followed by adjuvant hypo-fractionated or high dose single fraction SBRT. While CRC only represented 15 cases, the control rate was 93% ¹⁸.

Traditional classification schema of radioresistant versus radiosensitive pathologies were based on their response to cEBRT, which is delivered in one or two beams without precise conformal techniques. Consensus exists that tumor histology is perhaps the single most important factor in determining response to cEBRT ^20,21. In the review of 34 cases by Brown, et al. the local disease progression rate was 92.5%, with improved survival in those that received more than 30 Gy in 2-3 Gy per fraction ¹¹. The series by Katagiri, et. al. and Marzano, et. al. both demonstrated relatively poor response rates for radioresistant tumors at 33% and 20% respectively ^22,23. Image guided radiation therapy, including SRBT have demonstrated clinical responses of greater than 85% regardless of tumor histology. Yamada, et al. describe a case series of 811 lesions treated with SBRT in which 657 patients had radioresistant pathologies where dose was analyzed as a continuous variable ranging from 18-26 Gy single fraction (median dose of 1644 cGy in the low-dose group and 2240 cGy in the high-dose group). Local failure rates for the low-dose and high dose-groups were 5% versus 0.41% at 12 months, 15% versus 1.6% at 24 months and 20% versus 2.1% at 48 months, respectively. ¹²

Common genetic mutations in CRC include Kirsten-ras (K-ras), p53, and adenomatous polyposis coli (APC). In a large cohort by Smith, et. al. only 6.6% of tumors were found to have all 3 mutations, while 38.7% of tumors contained one of these genetic mutations. However, the most common combination of mutations was p53 and APC, 27.1% ²⁴. Further work by Wang, et. al. evaluated the prognostic value of mutations in the APC gene. In their cohort of 331 patients, APC-wild type gene mutation was present in 26% of patients, and portended poor overall survival in stage IV microsatellite stable CRC patients regardless of K-ras or BRAF status through multivariate analysis ²⁵. Of note, in our small series 3/7 patients who had progression of disease had APC exon 16 mutated, and the other 4/7 were not tested. APC mutation is an early, if not initiating event for 80-85% of sporadic CRC ²⁶. There is increasing evidence that mutations in the APC gene play critical roles in CRC tumorigenesis and progression of disease, which may be associated with a poor prognosis ²⁶. In our current series, we were unable to model the association of APC mutation with clinical outcomes due to the small number tested.

Our current analysis identified predictors of surgical outcome in this cohort. Patients presenting with neurologic deficits as described by ASIA scores or those with poor functional status as described via ECOG scores had worse overall survival. This is in line with previous reports ²⁷. In addition, patients with longer construct length may tend to fail more locally due to extensive local disease at presentation involving multiple levels. Finally, our data also shows that patients with lumbar disease have better local control. However, total dose and biologic equivalent dose (BED) were not associated with survival. These prognostic factors may be important for informed decision making and patient education, allowing reasonable treatment goals to be set.

Our study has several limitations, including the retrospective nature of this study. Second, due to the limited number of patients, in-depth statistical analysis of patients who had local failure was restricted. Additionally, complete genetic profiling was only acquired if it had a clinically significant impact on the management of systemic therapy. Further, because of the advent of novel therapies over the years, the population studied here was heterogenous in systemic treatments received ²⁸. Other confounding variables to take into consideration include the concomitant impact of systemic therapies on local disease control, the ability to complete treatment protocols based on patient tolerance and their impact on survival. Despite these limitations, the utilization of genomic sequencing highlights the growing role that it may play in prognostication of systemic disease control.

Conclusion

This study demonstrated that hybrid therapy (i.e., separation surgery followed by SBRT) in CRC patients presenting with MESCC resulted in sustained local control, with limited complications. Progression of spinal disease outside the treatment area is common and interval radiologic follow up may be useful for early detection. OS and PFS were not affected by mutational status, but APC mutations are commonly present in CRC patients with spinal metastases and may suggest worse prognosis. Patients with CRC spinal metastases commonly progress outside the index treatment and close clinical and radiographic follow up may lead to early identification of progression of disease and with timely radiation therapy, possible avoidance of further surgical interventions.

Figure 2:

Representative case. 52-year-old M presented with gait ataxia and back pain, found to have a T5 Lesion with MESCC. He underwent a T4-6 decompression and Instrumented fusion. A-B: Preop MRI. C: Postoperative X-ray. D-E: Postoperative CT-Myelogram. H-I: SBRT Plan with Isodose lines. Note that post-operative contours account for pre-operative tumor volume as per consensus guidelines.

Glossary

CRC

Colorectal cancer

OS

Overall Survival

PFS

Progression-free survival

MESCC

Metastatic epidural spinal cord compression

cEBRT

Conventional external beam radiation therapy

SBRT

Stereotactic body radiotherapy

GTV

Gross tumor volume

CTV

Clinical tumor volume

PTV

Planning target volume

PMMA

Poly-methyl-methacrylate

K-ras

Kirsten-ras

APC

adenomatous polyposis coli

BED

biologic equivalent dose