Clival chordoma research
Home › Treatment and outcomes

Outcome determinants

Agent report for Kya, 8 Sep 2026. Scenario: 34-year-old woman, 3.8 cm clival mass partly encasing the basilar, abutting both cavernous carotids, into the cavernous sinus; working frame dedifferentiated or poorly differentiated chordoma pending biopsy. Every PMID checked on PubMed E-utilities 8 Sep 2026; every figure is from the abstract unless marked UNVERIFIED. L1 trial; L2 meta-analysis or systematic review; L3 large series or registry; L4 small series; L5 judgement. HR above 1 is worse.

1. Histological subtype and tissue markers

  1. Chondroid vs conventional. Chondroid carries about half the progression risk (pooled HR 0.50, 95% CI 0.36 to 0.69). Zou 2018, PMID 29045855. L2.
  2. Dedifferentiated (high-grade sarcoma beside chordoma): under 1 percent of chordomas, median age 54; the sarcoma part loses brachyury and often carries TP53 mutation; 6 of 7 followed patients metastasised (Hung 2020, PMID 32427623, L4). Independent predictor of chordoma-specific death in 284 clival cases (Bai 2022, PMID 34623525, L3; Bakker 2018 review, PMID 30220042, L2).
  3. Poorly differentiated, SMARCB1 (INI1) deficient: median age 11 (range 1 to 29), 53 percent clival, brachyury kept, INI1 lost, survival significantly shorter than other subtypes after matching for site (Shih 2018, PMID 29483606, L4; Hasselblatt 2016, PMID 27067307). In 73 childhood skull base chordomas all but one poorly differentiated case died while the rest did well (Hoch 2006, PMID 16819322, L3; Beccaria 2018, PMID 29361006, L3). Caution: partial INI1 loss occurs in 32 percent of conventional skull base chordomas with no survival effect; diagnosis needs morphology plus complete loss (Righi 2023, PMID 37456229, L3).
  4. Ki-67. 5 percent or more meant shorter survival in 28 (Horbinski 2010, PMID 20670138, L4); under 6 percent independently favourable in 210 (Passeri 2023, PMID 36272118, L3); in the Pittsburgh prospective panel it dropped out once chromosome losses were counted (Zenonos 2018, PMID 29905508, L3).
  5. 1p36 loss and 9p21 (CDKN2A) homozygous deletion: the two independent predictors in 105 prospectively tested clival chordomas; strata of 1p36 (0 to 15 vs over 15 percent of cells) and 9p21 (0 to 3, 4 to 24, 25 percent or more) span cumulative HR 1 to 56 for progression after surgery and 1 to 76 after radiotherapy (Zenonos 2018, L3). CDKN2A loss is the commonest change acquired at progression (Cottone 2020, PMID 31916407, L3).
  6. PBRM1 plus 22q (SMARCB1) loss: HR 10.6 (2.8 to 39.6) for chordoma-specific death, 4.3 for recurrence, in 80 whole genomes (Bai 2021, PMID 33536423, L3).
  7. Brachyury: diagnostic, not prognostic (Zhang 2013, PMID 24086644, L3); its loss marks the dedifferentiated component.
  8. TP53: mutated in 4 of 5 dedifferentiated or poorly differentiated vs 1 of 102 conventional (OR 163; 39 vs 1.2 percent across the literature), acquired at progression, worse survival within the aggressive group (Baluszek 2025, PMID 41420192, L4). p53 staining alone is not prognostic (Horbinski 2010).

2. Extent of resection

  1. Meta-analysis, 807 patients. Incomplete resection: 3.8 times the recurrence (1.6 to 9.0) and 5.9 times the deaths (1.4 to 24.5) at 5 years; progression-free survival 20.7 points lower. Di Maio 2011, PMID 21819197. L2.
  2. Pooled HRs, 1,754 patients. Non-total resection HR 2.01 (1.54 to 2.62) for progression and 5.16 (2.27 to 11.70) for death; adjuvant radiotherapy HR 0.30 (0.16 to 0.56) for progression. Zou 2018, L2 (also Zou 2017, PMID 28437876).
  3. Large series agree. Beijing 238: marginal resection improved survival; recurrence and volume 40 cm3 or more predicted failing to get it (Wang 2017, PMID 28059654, L3). Beijing 242: gross total 35.5 percent, better on both endpoints (Zhai 2018, PMID 29303447, L3). Paris 210: gross total resection and Ki-67 under 6 percent the only independent predictors (Passeri 2023). Pittsburgh endonasal 60: gross total 67 percent, 83 in primary cases, 89 in later years (Koutourousiou 2012, PMID 22592328, L3). Also Sen 2010, PMID 19929198, L3.

3. Size, volume and spread

  1. Size. 4 cm or more: HR 4.03 for death in 736 National Cancer Database patients (Soffer 2024, PMID 38691856, L3; also Jones 2014, PMID 24852903, L3). Over 20 cm3 limited gross total resection (Koutourousiou 2012). For radiotherapy, over 25 ml: local control 56 vs 100 percent (Hug 1999, PMID 10470818, L3; Ares 2009, PMID 19386442, L3). The patient's 3.8 cm sits just under the 4 cm line; an irregular 3.8 cm mass is probably 15 to 30 ml, straddling the 20 to 25 ml cut-offs (estimate, L5; ask for the measured volume).
  2. Spread that limits resection. In 210 patients intradural extension, brainstem contact, cavernous sinus extension, major artery encasement, petrous apex extension and high volume each reduced extent of resection; prior surgery reduced it and raised CSF leak (Passeri 2023). Lower clivus with lateral extension and previous treatment were the limits in Pittsburgh and Beijing (Koutourousiou 2012; Bai 2022). A score from size, site, vascular encasement, intradural extension, brainstem invasion and recurrence predicted resection, stages, recurrence and postoperative performance (Brito da Silva 2018, PMID 29126120, L4). Dural penetration independently shortened survival in 67 (Zhai 2017, PMID 27506406, L4; not in 47, Rachinger 2014, PMID 24405075, L3).
  3. Spread that limits radiotherapy. Brainstem involvement forced dose reduction and cut control from 94 to 53 percent (Hug 1999); brainstem compression predicted failure (Ares 2009); optic or brainstem compression, histology and volume were the independent predictors of failure and death in 222 (Weber 2016, PMID 27247057, L3), and volume, compression and dose coverage in 135 (Iannalfi 2020, PMID 32193546, L3); over 3 mm from the brainstem meant better control (Mattke 2023, PMID 36149438, L3); contact worsened survival in 76 (Skóra 2026, PMID 42535416, L3). Whether the patient's tumour touches brainstem or optic apparatus is the question for the neuroradiologist.
  4. Prior treatment. Ten-year recurrence-free 42 percent for first operations vs 26 after reoperation (Tzortzidis 2006, PMID 16883163, L3); primary cases did better on carbon (Uhl 2014, PMID 24948519, L3). The patient is untreated: in her favour.

4. Radiotherapy dose and modality

  1. Dose. In 715 NCDB chordoma patients, 70 Gy or more improved 5-year survival (definitive 79 vs 54 percent; postoperative 83 vs 77; univariate); protons independently improved survival in the definitive setting. Palm 2019, PMID 30644538. L3. Standard prescriptions: 73.5 to 74 Gy(RBE) protons, 66 to 70.4 Gy(RBE) carbon.
  2. Modality. Protons and carbon gave similar control, survival and toxicity in 147 (Mattke 2023) and 135 (Iannalfi 2020); no 5-year difference between particles and modern photons in the meta-analysis (Di Maio 2011).
  3. How anatomy limits dose. Brainstem and optic constraints, not the prescription, set the dose to tumour touching them (Hug 1999; Weber 2016; Iannalfi 2020); hence surgeons separate tumour from those structures even when total removal is impossible (L5; Stacchiotti 2015, PMID 25638683).

5. Age and sex

  1. Age. Older age raises the hazard about 3 percent per year (pooled HR 1.03, Zou 2018; 1.028 in SEER, Jones 2014). Under 25 progressed less (Tian 2016, PMID 27237421, L3); under 48 did better on carbon (Uhl 2014); skull base tumours present youngest and survive longest of any site (Zuckerman 2018, PMID 29486315, L3). Children with conventional histology beat adults; those with poorly differentiated histology do very badly (Hoch 2006); in 141 children, age 10 or over had HR 0.26 for death and dedifferentiated histology HR 8.06 for cancer death (Clementi 2026, PMID 41252601, L3). At 34 the patient is on the good side of every adult cut-off; poorly differentiated chordoma is rare above 29 and dedifferentiated peaks at 54, so by age alone conventional is the likeliest histology (L5).
  2. Sex. Contradictory: male worse in 47 (progression-free 4.8 vs 9.8 years; Rachinger 2014); female worse 10-year control after protons (42 vs 65 percent; Munzenrider 1999, PMID 10394399, L3; Bakker 2018 agrees). Unresolved.

6. Fitness, performance status, frailty, nutrition

  1. Performance status. Each 10 Karnofsky points cut mortality 4.5 percent in 41 skull base chordomas (Roy 2021, PMID 34979650, L4); not prognostic in Rachinger 2014. Preoperative visual deficit (diplopia counts): pooled HR 2.77 for progression, probably marking spread (Zou 2018, L2).
  2. Frailty (cranial surgery generally; nothing chordoma-specific). Modified frailty index predicted death, complications and long stay in 9,149 oncological neurosurgery cases (Youngerman 2018, PMID 29139096, L3) and life-threatening complications after meningioma surgery (aOR 1.40 per stratum; Dicpinigaitis 2021, PMID 34495456, L3). In endonasal skull base surgery it predicted readmission (OR 2.35) but not complications, leak or death (Kshirsagar 2024, PMID 37746938, L4; Olsson 2025, PMID 40960267, L2; Qureshi 2023, PMID 37624530, L2). The patient scores zero; the indices miss weight loss and deconditioning.
  3. Albumin and weight. Albumin under 3.5 g/dl: OR 1.91 for 30-day death and 1.46 for non-routine discharge in 11,510 craniotomies; obesity raised major complications (OR 1.28 to 1.67) (Dasenbrock 2017, PMID 27203139, L3). Underweight was not separately analysed (gap). High BMI raises CSF leak (OR 1.61 per 5 kg/m2; Dlouhy 2012, PMID 22443502, L3); low BMI has no such signal. For the patient at 51 kg on opioids with poor intake, albumin and weight are the modifiable items; dietitian before surgery (thread 18, item 31).

7. Recovery time

  1. Thread 18 covers the timeline and its lengtheners. Additions: a complicated surgical course independently predicted worse progression-free and overall survival in 83 (Li 2026, PMID 42084902, L3); the Brito da Silva score ties the anatomy that predicts incomplete resection to more stages and lower postoperative performance; prior surgery raises leak at reoperation (Passeri 2023). The anatomy that blocks removal also lengthens recovery.

Ranked table

Rank Factor Direction Size of effect Level Modifiable
1 Aggressive histology Worse Largest; TP53 OR 163; paediatric HR 8 L3 to L4 No
2 Gross total resection Better Incomplete: death 5.2 to 5.9 times, recurrence 2 to 3.8 L2 Partly
3 1p36 loss, 9p21 deletion Worse Cumulative HR to 56 to 76 L3 No
4 Brainstem, optic, vessel, cavernous sinus, dural spread Worse Control 94 to 53 percent L3 No
5 Size 4 cm or 20 to 25 ml Worse HR 4.0; control 100 vs 56 percent L3 No
6 Radiotherapy 70 Gy(RBE) or more Better HR 0.30; 25-point survival gap L2 to L3 Yes
7 Prior surgery or radiotherapy Worse Recurrence-free 42 vs 26 percent L3 Yes
8 Ki-67 5 to 6 percent or more Worse Independent in 2 series L3 No
9 Older age Worse HR 1.03 per year L2 No
10 Low albumin, weight, conditioning Worse Death OR 1.9 (craniotomy) L3 Yes
11 Frailty indices Worse Stay, readmission only L2 to L3 Partly
12 Performance status Better 4.5 percent per 10 points L4 Partly
13 Sex Unresolved Opposite findings L3 No

Relative weight

Subtype comes first because it changes the disease rather than the odds: conventional chordoma is a slow local problem where surgery plus 74 Gy(RBE) buys years; dedifferentiated and poorly differentiated tumours metastasise and need chemotherapy-based treatment. Until biopsy, age 34 and a mass probably present ten months ago both point toward conventional. Within conventional chordoma the 1p36 and 9p21 losses and Ki-67 sort the spread of behaviour; they too need tissue. Extent of resection is the strongest treatment-side factor and the only large one anyone can act on. Spread and size act through resection and through the dose the brainstem and optic pathways allow, and the patient sits at the boundary on both; age is a modest effect in her favour. Fitness and nutrition do not change tumour biology but do change complications, which lengthen recovery and in one series shortened survival; they are the only items the patient and her partner can work on this week.

Absolute figures

Held back from this page.

Research thread 19 · source file em-research/19-outcome-determinants.md · last edited 08 September 2026