Follow-up
The median follow-up time after radiation treatment was 60.3 months (range 1.3 – 97.4 months). Overall, 72% of the patients were followed-up for at least 5 years or until death. The lost to follow-up rate after 1, 2 and 5 years amounted to 3% (2 patients), 7% (5 patients) and 13% (9 patients), respectively. The baseline characteristics of the study population are listed in table 1.
Table 1: Patients’ properties
Characteristics
|
Value
|
%
|
Age (years)
- Median
- Range
|
61
34 – 84
|
|
Gender (n)
- Female
- Male
|
22
46
|
32.4
67.6
|
Resection status (n)
- Biopsy
- R2
- R0/1
|
28
26
14
|
41.2
38.2
20.6
|
Treatment (n)
- Primary
- Recurrent
|
52
16
|
76.5
23.5
|
Most cranial level of tumor (n)
- L4/5
- S1
- S2
- S3
- S4
- S5
- Os coccygeum
|
7
12
23
14
5
4
3
|
10.3
17.7
33.8
20.6
7.4
5.9
4.4
|
GTV (ml)
- Median
- Range
CTV1 (ml)
- Median
- Range
CTV2 (ml)
- Median
- Range
PTV1 (ml)
- Median
- Range
PTV2 (ml)
- Median
- Range
|
182
0 – 1727
263
0 – 1743
938
60 – 2577
414
0 – 2325
1109
84 – 3138
|
|
Radiation dose ED 2 Gy
Carbon ion only: (α/β = 2)
- 60 Gy/3 Gy (RBE) 75.0 Gy
- 63 Gy/3 Gy (RBE) 78.8 Gy
- 66 Gy/3 Gy (RBE) 82.5 Gy
- 64 Gy/4 Gy (RBE) 96.0 Gy
IMRT + carbon ion boost
- 50 Gy/2 Gy + 15 Gy/3 Gy (RBE) 68.8 Gy
- 50 Gy/2 Gy + 18 Gy/3 Gy (RBE) 74.5 Gy
- 50 Gy/2 Gy + 24 Gy/3 Gy (RBE) 80.0 Gy
|
16
2
14
14
1
1
20
|
23.5
2.9
20.6
20.6
1.5
1.5
29.4
|
Abbreviations: GTV = gross tumor volume, CTV1 = clinical target volume (Boost plan), CTV2 = clinical target volume (primary plan), RBE = relative biological effectiveness, IMRT = intensity-modulated radiotherapy.
Local control
Local recurrence was observed in 31 patients (46%), occurring after a median follow-up time of 25.3 months (range 2.5 – 73.1 months). Most local recurrences were unifocal and occurred in the PTV2 region (24/31 cases, 77%); the remaining recurrences were either multifocal and originated in the PTV1 and PTV2 region (3/31 cases, 10%) or unifocal but separated from the primary tumor in the PTV1 region (4/31 cases, 13%). The majority of local relapses (22/31 patients; 71%) occurred within the first 3 years after primary treatment, but in 3 patients (10%) local relapse was observed later than 5 years after irradiation. For the overall study population, the 1-, 2-, 3- and 5-year probabilities for local tumor control were 90% (95% CI 81 – 95%), 80% (95% CI 70 – 89%), 65% (95% CI 54 – 77%) and 53% (40 – 67%), respectively.
The majority of patients with local relapses received systemic therapy, re-irradiation and/or surgical resection; only 4 patients decided for a wait-and-see approach. Amongst the tested prognostic factors, only the therapy situation before irradiation (primary vs. recurrent tumor) proved to be statistically significant for prediction of local tumor relapse (see table 2 and figure 2). For patients with a primary disease, local tumor control after 1, 2, 3 and 5 years was 96% (95% CI 88 – 99%), 88% (95% CI 78 – 95%), 77% (95% CI 65 – 88%) and 62% (95% CI 47 – 76%). In patients with recurrent chordomas, the 1-, 2-, 3- and 5-year LC rates reached only 68% (95% CI 44 – 89%), 54% (95% CI 31 – 80%), 27% (95% CI 10 – 60%) and 27% (95% CI 10 – 60%), respectively.
In contrast, the GTV, treatment approach (carbon ions only vs. combined photons and carbon ions) and applied cumulative equivalent radiation dose (EQD2) were statistically insignificant for LC in the Cox regression analysis (see table 2).
Table 2: Analysis of prognostic factors related to local control after RT.
|
Univariate testing
|
Multivariate testing
|
Parameter
|
p-value
|
HR
|
95% CI
|
p-value
|
HR
|
95% CI
|
EQD2 (Gy)
|
0.13
|
0.95
|
0.90 – 1.01
|
0.14
|
0.96
|
0.90 – 1.02
|
Therapy situation (primary vs. recurrent tumor)
|
< 0.001
|
4.25
|
1.99 – 9.09
|
< 0.001
|
4.55
|
2.06 – 10.08
|
GTV (ml)
|
0.59
|
1.21
|
0.60 – 2.46
|
0.16
|
1.70
|
0.81 – 3.57
|
Treatment approach (C12 vs. C12 + Ph)
|
0.49
|
1.29
|
0.63 – 2.64
|
NA
|
NA
|
NA
|
Abbreviations: HR = hazard ratio, 95% CI = 95% confidence interval, EQD2 = cumulative equivalent radiation dose, GTV = gross tumor volume, C12 = carbon ions, Ph = photons, NA = not analyzed
Distant metastases
Six patients (9%) developed distant metastasis, half of which were located in the pelvis. Extra-pelvic metastases manifested in the autochthonous back muscles of the thoracic and lumbar spine (2 patients) and in the lungs (1 patient). In all patients, local relapse was diagnosed prior to (2 patients) or concomitant to metastatic spread (4 patients). The median time until first distant failure of the disease was 32.5 months (range 12 – 67 months) after carbon-ion based RT. Distant metastases were treated with surgical resection and/or systemic therapy (e.g. tyrosin kinase inhibitors). The 1-, 2-, 3- and 5-year MFS rates were 91% (95% CI 64 – 100%), 81% (95% CI 54 – 97%), 71% (95% CI 45 – 93%) and 52% (95% CI 30 – 79%), respectively.
Survival
Twenty-three patients (34%) died during follow-up. In 17 of those 23 patients (74%), local relapse with or without additional distant metastases have been diagnosed before death. After a follow-up period of 1, 2, 3 and 5 years, the PFS rates were 90% (95% CI 81 – 95%), 80% (95% CI 70 – 89%), 65% (95% CI 54 – 77%) and 53% (40 – 67%), while the OS rates amounted to 97% (95% CI 93 – 100%), 97% (95% CI 93 – 100%), 86% (95% CI 77 – 95%), and 74% (95% CI 63 – 86%), respectively. Univariate log-rank tests found statistically significant associations of the patient’s age at diagnosis group, GTV group and the PTV2 (primary plan) group with OS after RT of sacral chordomas (p = 0.01, p = 0.02 and p = 0.0001), see figures 3 – 5. Further tested factors including the patients’ gender, the radiation dose, the localization of the tumor, the CTV1 (boost plan), CTV2 (primary plan), PTV1 (boost plan), disease situation (primary vs. relapsed disease) or the radiooncological therapy approach (carbon ion irradiation alone or combined irradiation with photons and carbon ions) were not statistically significant for survival prediction after RT in the log-rank test.
In univariate Cox models for continuous predictors, patients’ age at the beginning of RT, PTV1 and PTV2 were found to have statistically significant associations with OS after RT (see table 3). In the multivariate analysis, none of the factors tested showed statistical significance for prediction of OS (see table 3).
Table 3. Analysis of prognostic factors related to overall survival after RT.
|
Univariate testing
|
Multivariate testing
|
Parameter
|
p-value
|
HR
|
95% CI
|
p-value
|
HR
|
95% CI
|
Patients’ age (y)
|
0.01
|
1.055
|
1.013 – 1.10
|
NA
|
NA
|
NA
|
EQD2 (Gy)
|
0.48
|
0.971
|
0.895 – 1.05
|
0.533
|
0.975
|
0.899 – 1.06
|
Level of proximal invasion ≥ S2
|
0.21
|
1.765
|
0.725 – 4.30
|
NA
|
NA
|
NA
|
Recurrent tumor
|
0.31
|
1.593
|
0.654 – 3.88
|
0.36
|
1.539
|
0.628 – 3.77
|
GTV (ml)
|
0.06
|
1.001
|
1.00 – 1.00
|
NA
|
NA
|
NA
|
PTV1 (ml)
|
0.04
|
1.001
|
1.00 – 1.00
|
NA
|
NA
|
NA
|
PTV2 (ml)
|
0.002
|
1.001
|
1.00 – 1.00
|
NA
|
NA
|
NA
|
Abbreviations: HR = hazard ratio, 95% CI = 95% confidence interval, EQD2 = cumulative equivalent radiation dose, GTV = gross tumor volume, PTV1 = planning target volume (boost plan), PTV2 = planning target volume (primary plan), NA = not analyzed
Toxicity
At baseline, most patients had tumor- or operation-related complaints, with pain, sensitivity and bladder emptying disorders being the most common impairments; other initial complaints were motor deficits, rectal disorders and urinary and fecal incontinence (see table 4). The operated patients had considerably more neurological impairments than the patients without surgery (bladder emptying disorders: 48% vs. 29%, rectal disorders: 23% vs. 18%, urinary and fecal incontinence: 18% vs. 11%, sensitivity disorders: 35% vs. 21% and motor deficits: 8% vs. 0%). Furthermore, 6 patients had an enterostoma, all affected patients had a sacral chordoma at level S1 – 3 and 5 patients had a previous surgery.
After RT, 40 patients (59%) developed radiogenic late toxicities, which affected the bone and nerve tissues, the gastrointestinal tract and the skin (see table 4). The incidence of late toxicities ≥ grade 3 was 21% (14 of 68 patients). Of the 14 patients with severe radiogenic late toxicities, 12 received a radiation dose of at least 80 Gy (EQD2). Sacral insufficiency fractures (SIFs) were by far the most common late side effect in our analysis accounting for 49 % of the patients (33 of 68 patients). Of these, 36% (12 of 33 patients) were symptomatic with considerable impairments in everyday life and severe pain requiring multiple pain medications as well as intensive care by a pain specialist. The median time until diagnosis of SIFs was 12 months (range 1 – 62 months), whereby the majority of SIFs occurred within the first 2 years after RT (85%, 28 of 33 patients).
Radiogenic damage to the peripheral nerves, intestinal tract and skin was much less frequently observed with an incidence of 9% (6 patients), 3% (2 patients) and 9% (6 patients), respectively (see table 4).
Table 4: Late morbidities (CTCAE v5.0) after carbon-ion based radiotherapy for primary and recurrent sacral chordoma
Grade
|
0
|
1
|
2
|
3
|
4
|
Skin
|
62
|
6
|
0
|
0
|
0
|
Gastrointestinal tract
|
66
|
0
|
2
|
0
|
0
|
Peripheral nerves
|
62
|
1
|
2
|
1
|
2
|
Sacral insufficiency fractures
|
35
|
21
|
1
|
11
|
0
|
The values given are patient numbers.
Abbreviations: EQD2 = cumulative equivalent radiation dose, GTV = Gross tumor volume