Bone turnover biomarkers reflect radiation-induced bone injuries in women with non-metastatic rectal cancer

Abstract

Preoperative radiotherapy (RT) for non-metastatic rectal cancer reduces local recurrence rates but can cause pelvic insufficiency fractures. Despite the high morbidity from RT-induced skeletal injuries, predictive and preventive measures are lacking. How these injuries are reflected by bone biomarkers are largely unknown. The aim was to assess longitudinal changes in bone biomarkers and their relation to RT-related bone injuries in women with rectal cancer. This longitudinal cohort study includes 47 women with non-metastatic rectal cancer treated with surgery ± preoperative RT with or without chemotherapy. Sclerostin, bioactive sclerostin, C-terminal telopeptide cross-links of collagen type I (CTX), bone-specific alkaline phosphatase (BALP), and type I procollagen intact N-terminal propeptide (PINP) were measured at baseline, after RT, and 1 yr postoperatively. Pelvic magnetic resonance imaging was used for detection of skeletal injury. Sixteen of 36 (44%) irradiated women had radiation-induced bone injuries and were compared to 11 women (RT–) and 20 women (RT+) without bone injuries. Serum CTX, BALP, and PINP increased during the first year after RT in women with radiation-induced bone injuries. The difference in mean change of CTX (p=.037) and BALP (p=.042) was conferred by longitudinal regression analyses adjusted for serum estradiol. Serum sclerostin and bioactive sclerostin remained stable over time. Taken together, bone markers may be of interest for future research on fracture prediction or preventive measures in women susceptible to radiation-induced bone injury. Due to few measure points, the full pattern cannot be captured regarding the relation over time between bone biomarkers and skeletal injury from irradiation.

Introduction

The majority of patients with non-metastatic rectal cancer are treated with preoperative radiotherapy (RT),^1,2 since the local recurrence rate is reduced by about 50% in comparison with surgery alone.³ Recent studies report a complete tumor response from RT combined with chemotherapy in selected individuals,^4–6 in whom extensive surgery can be avoided.^7,8 Accordingly, RT has gained an increasing interest with an expected growing number of irradiated rectal cancer survivors, emphasized by the steadily rising incidence of early onset colorectal cancer in high-income countries.^9,10

Although effective on the tumor, RT comes with troublesome adverse toxic effects on the surrounding healthy tissue that is exposed to the radiation field. Pathophysiological injury from irradiation includes early red bone marrow differentiation into fat cells,¹¹ late vascular injury,¹² microarchitectural changes in corticoid and trabecular bone,^13–15 with osteocyte loss¹⁶ and damage to osteoblasts with a subsequent decrease in collagen production resulting in impaired bone quality.^12,17 A clinical research goal is to identify factors that predict and diagnose RT-induced skeletal lesions. Clinical risk factors have been linked to higher age, post-menopausal status, corticosteroid therapy, and diabetes mellitus, that is, risk factors similar to post-menopausal and age-related fractures.^18,19 Bone biomarkers have been suggested as potential surrogate markers, in addition to radiographic imaging, to report on altered bone properties and the presence of insufficiency fractures associated with pelvic RT.^18,20 However, further studies are needed to understand the response of bone biomarkers in relation to the involved bone cells and bone injuries after RT. Biochemical markers of bone turnover are useful for monitoring anabolic and antiresorptive treatments for osteoporosis, to assess compliance to therapy, and possibly to predict risk of fractures and identify secondary osteoporosis.^21,22

Three different types of bone cells produce and maintain bone through the remodeling process that can be assessed by biochemical markers of bone turnover.²² Osteoclasts resorb mature bone tissue and can be assessed by the bone degradation product C-terminal telopeptide cross-links of collagen type I (CTX). Osteoblasts synthesize new osteoid matrix and are crucial for adequate mineralization, reflected as type I procollagen intact N-terminal propeptide (PINP) and bone-specific alkaline phosphatase (BALP). Osteocytes, the most abundant bone cells, orchestrate bone remodeling in response to both hormonal and mechanical cues.²³ Mature osteocytes produce sclerostin, which is a pivotal regulator of bone formation through its inhibitory effect on the Wnt signaling pathway. Sclerostin is not considered a conventional marker of bone turnover since it is a regulatory molecule rather than a product from the resorption and formation processes. Several immunoassays have been developed for the measurement of circulating sclerostin, and second-generation assays for intact sclerostin (a.k.a. bioactive sclerostin) have recently become available that might improve the clinical significance of sclerostin as a bone biomarker.²³

Although it is well established that RT has detrimental effects on bone tissue metabolism, leading to subsequent bone injuries, it is not understood how this is reflected by bone biomarkers.^17,20 The primary aim of the current study was to assess baseline and longitudinal changes in bone biomarkers in a cohort of women with and without RT-related bone injuries detected by MRI. The secondary aim was to investigate a possible association between bone biomarkers and bone injury, which might be of interest for the evaluation of risk for insufficiency fractures after preoperative RT.

Materials and methods

Study participants and design

The original multi-center cohort study included 139 women treated with surgery for non-metastatic rectal cancer to compare adverse effects of preoperative RT on sexual function,²⁴ ovarian androgen production,²⁵ and bone biomarkers.²⁶ In the latter study, RT was associated with an increase in PINP during the first year after cancer treatment. For many participants of the irradiated group, the radiologic follow-up included only CT scans, which are inferior to MRI for detection of radiation-induced insufficiency fractures.^27–30 The current study is restricted to 50 women included at Ersta Hospital (Stockholm, Sweden) as this center performed pelvic MRI in addition to CT of the thorax, abdomen, and pelvis for follow-up of rectal cancer. Magnetic resonance imaging was used to compare changes in bone biomarkers between women with and without radiological signs of radiation-induced bone injuries.

Inclusion criteria were women with non-metastatic rectal cancer, planned for abdominal resection of the rectum ± preoperative RT with or without oral capecitabine. Recommendations regarding oncological regimens and surgical procedures were made at multidisciplinary team conferences dependent on patient and tumor characteristics. Metastatic disease, inability to give informed consent, and treatment with bisphosphonates were exclusion criteria.²²

Preoperative RT was delivered with so-called box-technique, that is, 3 or 4 fields of photon irradiation (anterior–posterior, posterior–anterior, left–right, right–left), either as short-course (5 × 5.0 Gray (Gy)) or long-course (25–28 × 1.8–2.0 Gy) RT, comprising the regional lymph nodes at risk of subclinical disease. The surrounding pelvic tissue is to a large extent exposed to irradiation.³¹ Following RT, selected women underwent either “immediate” surgery within a week or “delayed” surgery after a waiting time of 6–8 wk. Surgical procedures were anterior resection of the rectum with or without a colorectal anastomosis, or abdominoperineal excision of the rectum. Selected women were treated with adjuvant oral capecitabine or combination chemotherapy with 5-fluorouracil and oxaliplatin for up to 6 mo postoperatively.

Fasting venous blood samples were collected at baseline before start of any rectal cancer treatment, on the day before surgery (only participants with preoperative RT), and 1 yr postoperatively. The change in bone biomarkers was compared between women with and without signs of radiation-induced bone injuries on MRI within 3 yr. Non-irradiated women were analyzed in the group without radiation-induced bone injury. Variables regarding clinical and pathological characteristics as well as data on treatment and outcome were longitudinally collected in a separate database and validated against medical records. This study was approved by the Regional Ethics Committee in Stockholm (2008/247-31/3, 2012/1730-32/3, 2020-02148).

Radiologic imaging

Magnetic resonance imaging of the lower abdomen and pelvis, and CT of the abdomen and chest were performed at baseline, 4–6 wk after RT before surgery in women with delayed surgery, and annually for 3 yr postoperatively.

Magnetic resonance imaging was performed by using a 3 Tesla MRI scanner (Verio, Siemens Healthcare, Erlangen, Germany). A standard protocol for rectal cancer was used, without contrast agent. Sequences include T1-weighted images in the transverse plane and T2-weighted images in all 3 dimensions. Coronal and transverse planes were angulated perpendicular to the tumor axis. The section thickness was 4–5 mm in the transverse plane and 3–5 mm in the sagittal and coronal plane, with a 0–0.5 mm interscan gap. The upper border was cranial to the intervertebral disc between the lumbar vertebra 5 and sacrum, and the lower border caudal to the soft tissue covering tuber ischii. The images were scrutinized for signs of local tumor recurrence and metastases as well as radiation-induced changes in bone. The latter were classified as insufficiency fractures, osteoradionecrosis in cortical bone, and signal changes in bone marrow. Insufficiency fractures were diagnosed in women with a fracture line displayed with low signal intensity in T1-weighted images. Fractures were combined with bone marrow edema, detected by low signal intensity in T1 and high in T2-weighted images. Signal changes in bone marrow have a typical appearance with an initially mottled heterogeneous pattern, usually detectable early, from 3 wk post-irradiation. This represents the gradual replacement of hematopoietic tissue by fat, which may proceed to a more advanced fatty replacement with a homogenous high signal.^32–34 All radiological examinations were reviewed by 2 MRI-radiologists with more than 20 yr of experience.

Biochemical measurements

Fasting blood samples were drawn between 08.00 and 10.00 h, and serum aliquots were stored at –20°C until analysis. Serum intact PINP was assessed with the UniQ radioimmunoassay (Aidian Oy), with an assay performance of: analytical range 5 to 250 μg/L, intra-assay coefficient of variation (CV) of <5%, and interassay CV of <6% at 53 μg/L. Serum BALP was measured by the MicroVue BAP quantitative ELISA (Quidel Corp.), with an assay performance of: analytical range 0.7 to 140 U/L, intra-assay CV of <6%, and interassay CV of <8% at 12 U/L. Serum CTX was assessed by the serum CrossLaps® ELISA (Immunodiagnostic Systems Holdings PLC), with an assay performance of: analytical range 20 to 3380 ng/L, intra-assay CV of <6%, and interassay CV of <10% at 450 ng/L. Serum sclerostin and bioactive sclerostin were measured by quantitative ELISA (Biomedica). The sclerostin assay (BI-20492) had an assay performance of: analytical range 3.2 to 240 pmol/L, intra-assay CV of <7%, and interassay CV <10% at 32 pmol/L. The bioactive sclerostin assay (BI-20472) had an assay performance of: analytical range 1.9 to 320 pmol/L, intra-assay CV of ≤2%, and interassay CV of ≤5% at 19 pmol/L. Serum estradiol was determined by radioimmunoassay (Estradiol-Ultrasensitive-Coated tube RIA, Cisbio Bioassays), with an assay performance of: analytical range 5 to 2000 pmol/L and CVtot of 16% at 57 pmol/L. All manual assays were performed in full accordance with the manufacturers’ instructions, and all samples were run in duplicates. Follicle-stimulating hormone (FSH) was measured by immunofluorescence using the PerkinElmer AutoDELFIA® system (PerkinElmer). The CVtot for FSH was 3% at 7.6 IU/L, 4% at 22 IU/L and 38 IU/L. Serum albumin and creatinine were analyzed using the Cobas 8000 instrument (Roche Diagnostics Scandinavia AB) by routine clinical laboratory assays at the Department of Clinical Chemistry (Swedac accredited no. 1886), Karolinska University Hospital, Sweden, with all intra-assay and interassay CVs <7%.

Statistical analysis

Statistical analyses were performed with Stata/IC 14.2 (StataCorp LLC) with an alpha error set at 0.05. The change in bone markers within groups was tested by the Wilcoxon matched-pairs signed rank test. Longitudinal data were analyzed with general equation estimation (GEE)-based regression models including the interaction between time point of bone marker measurement and group assignment (no bone injury vs bone injury). Final models were displayed graphically by margin plots. Adjustment for serum estradiol was preferred over menopausal status and FSH as participants could be pre-menopausal or treated by hormonal replacement therapy. Adjustment for the elapsed time between the end of RT and surgery, as well as for chemotherapy, minimally changed the estimates and was omitted.

Results

Three of 50 women were excluded from the current study due to treatment with bisphosphonates (n = 1) or loss to follow-up (n = 2). Clinical and treatment-related characteristics of the 47 participants are presented in Table 1. Median age for the total study group was 62 yr (26–90 yr). Results are displayed separately for the group of 16 women with subsequent radiation-induced bone injuries and the group of 31 women without bone injuries. RT treatment was given to all women with subsequent injuries and to 20 (65%) of women without bone injury. There were no significant differences between the groups (ie, with or without radiation-induced bone injuries) regarding age, BMI, American Society of Anesthesiologists classification of physical status, smoking status, menopausal status, chemotherapy treatment, type of surgery or tumor stage (Table 1).

Table 1.

Patient, tumor, and treatment-related characteristics for the study cohort.

	Radiation-induced bone injury (n = 16)	No bone injury (n = 31)	p-value
Age (yr)	62 (46–77)	61 (26–90)	.884a
BMI (kg/m2)	23.3 (18.6–35.9)	25.5 (19.3–36.1)	.426a
missing	1	0
ASA classification
I	6	7	.256b
II	9	16
III	1	8
Smokingc
Yes	3	1	.196b
No	8	16
Former	5	14
Menopausal at baseline			1.000b
Yes	13	26
No	3	5
Menopause 1 yr postop			.340b
Yes	15	31
No	1	0
Preoperative RT			NA
No RT	0	11
Short-course RT (5 Gy × 5)	11	15
Long-course RT (2.0 Gy × 25 or 1.8 Gy × 28)	5	5
Preoperative chemotherapyd			.416b
Yes	4	4
No	12	27
Postoperative chemotherapye			1.000b
Yes	3	7
No	13	24
Type of surgery			.696b
Anterior resection of rectum	10	15
Abdominoperineal excision of rectumf	6	15
Local excision of rectal cancer	0	1
Surgical approach			.455b
Laparoscopic	2	8
Open	14	22
pTNM
0–I	7	9	.636b
II	4	6
III	5	15
IVg	0	1

Abbreviations: ASA, American Society of Anesthesiologists classification of physical status; Gy, gray; pTNM, pathological tumor-node-metastasis staging; RT, radiotherapy.

NA = not applicable since radiation-induced bone injury only affects irradiated participants.

Values are shown as median (minimum–maximum) for continuous variables and numbers for categorical variables.

^a p-values calculated with Wilcoxon rank sum test.

^b p-values calculated with Fisher’s exact test.

^cYes, present or former <1 yr before inclusion; No, never; Former, >1 yr before inclusion.

^dOral capecitabine.

^eOral capecitabine (6 pts) or combination chemotherapy with 5-fluorouracil and oxaliplatin (4 pts).

^fIncluding one Hartmann’s procedure.

^gDistant metastases shortly after inclusion.

Radiological follow-up with MRI or CT (n = 41) or CT only (n = 6) was available in all individuals at least 1 yr postoperatively. Radiation-induced pelvic bone injuries were present in 16 (44%) of 36 irradiated women (Table 2). Changes were classified as insufficiency fractures with fracture lines in cortical bone (n = 11), osteoradionecrosis (n = 1), and signal changes in bone marrow (n = 15). Bone-related changes were detected between 1 mo after RT and 2 yr postoperatively.

Table 2.

Characteristics of radiation-induced injuries to cortical bone and bone marrow.

Patient no.	Age (yr)	Type of RT	Time from radiation to diagnosis of first injury	Cortical bone injury	Signal changes in trabecular bone and/or bone marrow	Localization and description a
1	71	5 × 5 Gy	1 yr	✓	✓	Bilateral, vertical sacral fractures. Unilateral pubic rami sup and inf fractures
2	62	5 × 5 Gy	1 yr	✓	✓	H-shaped sacral fracture (bilateral vertical and horizontal fracture lines). Bilateral pubic rami sup and inf fractures
3	69	25 × 2 Gy	1 yr	✓	✓	H-shaped sacral fracture (bilateral vertical and horizontal fracture lines)
4	59	5 × 5 Gy	1.5 yr	✓	✓	Bilateral, vertical sacral fractures
5	76	5 × 5 Gy	1 yr	✓	✓	Bilateral, vertical sacral fractures
6	61	5 × 5 Gy	1 yr	✓	✓	Bilateral, vertical sacral fractures
7	51	5 × 5 Gy	2 yr	✓	✓	Bilateral, vertical sacral, and iliac fractures
8	69	5 × 5 Gy	2 yr	✓	✓	Bilateral, vertical sacral, and iliac fractures
9	66	5 × 5 Gy	1 yr	✓	✓	Unilateral, vertical sacral fracture
10	63	5 × 5 Gy	1 yr	✓	✓	Unilateral, vertical sacral fracture
11	48	25 × 2 Gy	2 yr	✓	–	Horizontal, sacral fracture
12	46	25 × 1.8 Gy	1 yr	✓	✓	7 mm osteoradionecrosis in corticalis of os ilium and unilateral signal changes in sacrum
13	53	25 × 2 Gy	1 mo	–	✓	Bilateral signal changes in sacrum and os ilium
14	47	25 × 1.8 Gy	1.5 yr	–	✓	Bilateral signal changes in sacrum and os ilium
15	64	25 × 1.8 Gy	1 mo	–	✓	Bilateral signal changes in sacrum and os ilium
16	77	5 × 5 Gy	1 yr	–	✓	Unilateral signal changes in acetabulum

Abbreviations: Gy, gray; RT, radiotherapy.

^a Fractures in cortical bone and signal changes in sacrum and os ilium are parallel to sacroiliacal joints.

Medication with potential influence on bone metabolism

Treatments with drugs with a possible influence on bone metabolism are presented in Table 3. In total, 4 women had postoperative treatment with betamethasone for chemotherapy-induced nausea, including 2 women without and 2 with skeletal injury (woman number 2, injury detected 4 mo after betamethasone treatment; and woman number 13, injury detected before betamethasone treatment (Table 2)). One woman had only one dose of betamethasone 9 mo before and the remaining 3 stopped the treatment between 6 and 4 mo before the 1-yr follow-up sampling. No women had been treated with anti-osteoporosis medications.

Table 3.

Use of drugs in the total cohort with possible impact on bone metabolism.

	Baseline N	1-yr postop N
Betamethasone	0	4a
Menopausal hormone therapy	3	4
Oral contraceptives	1	0
Prednisolone (2.5–5 mg/d)	1	2
Calcium carbonate	4	5
Cholecalciferol	3	3
Metformin	2	2
Insulin	2	2
Levothyroxine	3	3

Abbreviation: N, number of patients.

^aBetamethasone treatment for chemotherapy-induced nausea, stopped 4–9 mo before the 1-yr follow-up.

Biochemical assessments

Results of the bone biomarkers, estradiol, FSH, creatinine, and albumin analyses are presented in Table 4. The baseline levels were comparable according to cross-sectional tests and longitudinal regression models between both groups (p>.30). Serum levels of CTX (p=.030), BALP (p=.016), and PINP (p =.019) increased during the first year after RT in women with radiation-induced bone injuries and remained stable in women without radiation-induced bone injuries. Serum levels of sclerostin and bioactive sclerostin were rather constant throughout the study period in both groups and within the reported reference interval with exception of one participant with decreased sclerostin and normal bioactive sclerostin. The difference in mean change of CTX (p=.037) and BALP (p=.042) between the 2 groups during the first year after cancer treatment was conferred by the longitudinal GEE regression analyses adjusted for serum estradiol and displayed in Table 5 and Figure 1. Adjustment of the final models of CTX and BALP with elapsed time between end of RT and blood sampling for the second study visit resulted in minimal changes of point estimates. The GEE models for PINP, sclerostin, and bioactive sclerostin were omitted due to non-significant global Wald tests, indicating that inclusion of further variables besides the baseline value did not improve fit of the model.

Table 4.

Comparison of bone biomarkers, estradiol, and FSH for women with and without radiation-induced bone injury.

	Radiation-induced bone injury			No bone injury
	N	Serum levels	Longitudinal comparisona p-valueb	N	Serum levels	Longitudinal comparisona p-valueb
Bioactive sclerostin (pmol/L)
Baseline	14	71 (43–100)		26	82 (22–184)
After RT	16	73 (40–155)		18	71 (30–181)
1 yr	14	67 (41–111)	.422	29	77 (28–236)	.300
Sclerostin (pmol/L)
Baseline	14	35 (21–50)		26	32 (3.2–84)
After RT	16	34 (16–76)		18	32 (16–75)
1 yr	14	33 (23–64)	.184	29	33 (3.2–92)	.518
CTX (ng/L)
Baseline	14	150 (60–350)		26	165 (50–450)
After RT	16	170 (100–330)		18	140 (50–320)
1 yr	14	215 (110–570)	.030	29	170 (70–660)	.747
BALP (U/L)
Baseline	14	12.0 (5.2–17.2)		26	10.4 (5.1–24.8)
After RT	16	12.5 (7.9–25.6)		17	10.2 (6.1–24.8)
1 yr	14	16.0 (10.0–30.8)	.016	29	12.1 (6.0–22.6)	.080
PINP (μg/L)
Baseline	14	48 (11–78)		26	57 (20–106)
After RT	16	50 (28–149)		18	53 (16–157)
1 yr	14	62 (38–143)	.019	29	57 (25–173)	.443
Estradiol (pmol/L)
Baseline	16	68 (29–343)		30	72 (33–2000)
After RT	16	49 (10–149)		20	68 (29–314)
1 yr	16	46 (23–88)	<.001	30	55 (10–180)	.007
FSH (IU/L)
Baseline	16	50 (1.8–82)		30	52 (1.1–98)
After RT	16	50 (4.6–104)		20	54 (6.6–101)
1 yr	16	52 (4.2–125)	.044	30	56 (33–94)	.162
Creatinine (μmol/L)
Baseline	16	59 (50–114)		31	67 (43–109)
After RT	16	60 (48–106)		23	63 (44–115)
1 yr	14	64 (38–113)	.046	26	68 (38–126)	.706
Albumin (g/L)
Baseline	16	44 (34–49)		31	43 (36–51)
After RT	16	38 (32–46)		20	42 (30–48)
1 yr	16	41 (35–45)	.038	29	41 (32–47)	.005

Abbreviations: BALP, bone-specific alkaline phosphatase; CTX, C-terminal telopeptide cross-links of collagen type I; FSH, follicle-stimulating hormone; N, number of patients; PINP, type I procollagen intact N-terminal propeptide; RT, radiotherapy.

Values are shown as median (minimum–maximum).

^aLongitudinal measures are calculated as baseline vs 1 yr after surgery.

^bWilcoxon signed rank test.

Table 5.

GEE regression analysis comparing change in bone biomarkers between participants with and without bone injury.

	Time period	Difference in CTX (ng/L)	Difference in BALP (U/L)
Unadjusted analysis	Short	48 (–23 to 120) p =.182	1.7 (–2.1 to 5.6) p=.380
	Long	67 (–1 to 135) p=.054	3.8 (0.0 to 7.5) p=.049
Adjusted for serum estradiol	Short	54 (–17 to 126) p=.136	1.9 (–2.0 to 5.8) p=.346
	Long	73 (4 to 141) p=.037	3.9 (0.1 to 7.7) p=.042

Abbreviations: BALP, bone-specific alkaline phosphatase; CTX, C-terminal telopeptide cross-links of collagen type I; RT, radiotherapy.

Short time period: from baseline to the day before surgery, which implies 0–79 d after the last day of preoperative RT.

Long time period: from baseline to 1 yr after surgery.

The estimates represent the difference (95% CI) in mean change of bone biomarkers over time between participants with and without bone injury (reference group: no bone injury).

Figure 1.

Estimated mean CTX and BALP with 95% CI derived from longitudinal regression models adjusted for serum estradiol illustrating the difference in mean change between women with vs without radiation-induced bone injuries. CTX, C-terminal telopeptide cross-links of collagen type I; RT, radiotherapy; BALP, bone-specific alkaline phosphatase.

Discussion

This longitudinal cohort study explored bone biomarkers in women with and without MRI-detected bone injury from preoperative RT for non-metastatic rectal cancer. Sixteen (44%) of the irradiated women had RT-induced bone injuries and increased bone biomarkers. In the regression analysis, adjusted for serum estradiol, changes in CTX and BALP were associated with radiation-induced bone injury. This is new knowledge that may be of interest for risk evaluation and preventive measures in women susceptible to radiation-induced bone injury.

RT, with curative intent or as neoadjuvant therapy before surgery, is an established treatment for rectal, anal, cervical, and prostate cancer although associated with negative side effects including damage to bone tissue and bone marrow.³⁵ Weakened bone after RT is susceptible for insufficiency fractures caused by normal stress.¹² Symptoms include pelvic and back pain, and immobility, which can affect quality of life substantially.³⁵ Reported incidences vary largely depending on, inter alia, method for definition and detection of fractures. MRI, which is a more sensitive method in comparison with conventional CT,^27–29 detected pelvic insufficiency fractures in 34% of irradiated rectal cancer patients 3 yr postoperatively in a nationwide Danish study.²⁸ This finding coincides with the current study where 16 women (44%) had radiation-induced injuries among whom 12 (33%) had fractures in pelvic cortical bone. Today, it is not possible to predict the individual risk of insufficiency fractures, but there are some established risk factors including radiation dose¹³ and technique³⁶ as well as patient characteristics, that is, high age, female sex, postmenopausal status, low BMI, osteoporosis,^35,36 and glucocorticoid use.^15,34 These risk factors were all evenly distributed between groups with and without radiation-induced bone injury in the present cohort. Bone biomarkers are of potential predictive use but the information in the literature is limited.³⁷ According to a Cochrane analysis from 2018, there is a significant lack of evidence regarding preventive measures of radiation-induced pelvic bone injury.¹⁸ These authors identified a need for interventional trials including individuals planned for pelvic irradiation, applying bone turnover markers as surrogate markers of bone health through prospective, and repeated sampling during treatment and follow-up, in addition to radiology and BMD measurements.¹⁸

In our previous study, focusing on the impact of RT on bone biomarkers in a larger cohort including the current,²⁶ RT was associated with an increase only in PINP during the first postoperative year, interpreted as a possible recovery process following bone injury. Encouraged by this new finding, further analyses were made and presented here, regarding the association between bone biomarkers and radiation-induced skeletal injury. In addition, bioactive sclerostin was added to the serum analyses. Serum levels of sclerostin and bioactive sclerostin remained, however, quite stable over time and between the 2 groups with and without skeletal lesions. This was somewhat surprising since it has been reported, in animal and in vitro models, that osteocytes are highly sensitive to radiation exposure since it induces osteocyte apoptosis.^37,38 However, neither sclerostin nor expression of the SOST gene (encodes sclerostin), was investigated in these studies. Experimental murine models with PTH 1–34 treatment and neutralization antibodies against sclerostin have shown that both these anabolic therapies alleviates radiation-induced bone loss through the Wnt signaling pathway.^39,40

Regarding the assessed bone turnover markers, significant differences were found between the groups. In the unadjusted analyses (Table 4), both resorption (CTX) and formation (PINP, BALP) markers increased over time in the group with bone injury but not in the group without; however, an increase in BALP was nearly significant in the latter. After adjustment in longitudinal GEE regression analyses (Table 5), increases in CTX and BALP were associated with radiation-induced bone injury. The findings regarding CTX and BALP may be of future interest in fracture prediction, or in selecting patents for prophylactic treatment of radiation damage to bone. This is not routinely performed today, but the concept of prophylactic treatment and the relation with bone turnover markers is under investigation in a randomized controlled trial from the United Kingdom, including individuals with cervical cancer. Intervention with oral bisphosphonates, in addition to calcium and vitamin D, is used in individuals considered at high-risk for radiation-induced bone injury.⁴¹ Bone turnover markers are measured as a secondary outcome. Results from the current study may be applicable for other pelvic cancers. Some clinical evidence suggests that intravenous zoledronic acid administered before irradiation of skeletal metastases may have a protective effect against radiation-induced bone toxicity.^42,43

Strengths and limitations

The analyzed study population is younger and healthier than the general population of women with rectal cancer since the original cohort study focused on sexual function, and participants of the current study were included at a center not treating locally advanced rectal cancer. However, the reported findings are important for a younger population in the light of the worldwide increasing incidence of early onset rectal cancer and expanding use of RT in early cancer stages for organ preservation. The venous blood samples were collected prospectively and analyzed in batches to reduce variability and information bias. The reported intervals in time (baseline, after RT, and 1 yr) are long and future studies should include more frequent and longer follow-up to enhance biochemical traceable changes of radiation-induced bone injuries, repair, and remodeling. Confounding by age, menopausal status, and estrogen treatment was assessed in the longitudinal regression analysis, and adjustment for serum levels of estradiol seemed to account for these confounders. A limitation of this study is that we only have data on radiation doses to tumors and not data on the absorbed dose to bones at different locations.

Conclusions

In this prospective exploratory study, bone turnover markers increased during the first postoperative year in women undergoing curative treatment for non-metastatic rectal cancer with adverse radiation-related side effects on bone detected with MRI. In the adjusted analyses, an association was found between the increase in CTX and BALP and radiation-induced bone injury. The findings may be of interest for further studies evaluating fracture risk and selection for prophylactic treatment. Due to few measure points, it was not possible to describe the relation in time between laboratory data and radiological findings.

Author contributions

Per Magnusson (Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Writing—original draft, Writing—review and editing), Maria Sääf (Conceptualization, Investigation, Methodology, Writing—original draft, Writing—review and editing), Anna Martling (Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Resources, Supervision, Writing—review and editing), Annika Svanström Röjvall (Conceptualization, Data curation, Formal analysis, investigation, Methodology, Validation, Writing—original draft, Writing—review and editing), Diana Atanasova (Conceptualization, Data curation, Investigation, Validation, Writing—review and editing), Franciszek Wilamowski (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Writing—review and editing), Angelique Flöter Rådestad (Conceptualization, Funding acquisition, Investigation, Resources, Writing—review and editing), Christian Buchli (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing—original draft, Writing—review and editing) and Josefin Segelman (Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Writing—original draft, Writing—review and editing)

Funding

Financial support was provided through the Swedish Cancer Society, the Stockholm Cancer Society, the Bengt Ihre Research Fellowship, the Bengt Ihre Foundation, and ALF grants from the Stockholm County Council, Karolinska Institutet, and Region Östergötland.

Conflicts of interest

P.M., M.S., A.M., A.S.R., D.A., F.W., A.F.R., C.B., and J.S. report no conflicts of interest.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.