Skip to main content

30-day mortality in patients treated for brain metastases: extracranial causes dominate



Established prognostic models, such as the diagnosis-specific graded prognostic assessment, were not designed to specifically address very short survival. Therefore, a brain metastases-specific 30-day mortality model may be relevant. We hypothesized that in-depth evaluation of a carefully defined cohort with short survival, arbitrarily defined as a maximum of 3 months, may provide signals and insights, which facilitate the development of a 30-day mortality model.


Retrospective analysis (2011–2021) of patients treated for brain metastases with different approaches. Risk factors for 30-day mortality from radiosurgery or other primary treatment were evaluated.


The cause of death was unrelated to brain metastases in 61%. Treatment-related death (grade 5 toxicity) did not occur. Completely unexpected death was not observed, e.g. accident, suicide or sudden cardiac death. Logistic regression analysis showed 9 factors associated with 30-day mortality (each assigned 3–6 points) and a point sum was calculated for each patient. The point sum ranged from 0 (no risk factors for death within 30 days present) to 30. The results can be grouped into 3 or 4 risk categories. Eighty-three percent of patients in the highest risk group (> 16 points) died within 30 days, and none survived for more than 2 months. However, many cases of 30-day mortality (more than half) occurred in intermediate risk categories.


Extracranial tumor progression was the prevailing cause of 30-day mortality and few, if any deaths could be considered relatively unexpected when looking at the complete oncological picture. We were able to develop a multifactorial prediction model. However, the model’s performance was not fully satisfactory and it is not routinely applicable at this point in time, because external validation is needed to confirm our hypothesis-generating findings.


Most patients with brain metastases from extracranial primary tumors such as lung or breast cancer receive palliative treatment approaches, because the common pattern of polymetastatic spread may cause compromised performance status (PS) and eventually also limited survival, often in the range of 3–9 months [1]. Both, longer survival (typically if oligometastases are present [2]) and shorter survival may be observed, and considerable efforts have been undertaken to predict survival (nomograms, scores, online calculators [3,4,5]). Given that very short survival often is synonymous to active treatment in the last 30 days of life, oncologists can opt for palliative and supportive care rather than brain-directed therapy [6]. Supposing they choose brain-directed therapy, the challenge is to navigate a complex scenario of low-value care, potential overtreatment and futile, but costly procedures [7, 8].

Prognostic models familiar to many providers, such as the diagnosis-specific graded prognostic assessment (DS-GPA) [3], were not designed to specifically address very short survival. Different definitions of very short survival may be applied, including 30-day mortality, which has been evaluated in numerous oncology settings [9, 10]. General survival prediction models such as TEACHH may also be utilized [11, 12], but it is still unclear whether a brain metastases-specific 30-day mortality model should be preferred. Our group has previously studied different models that predict very short survival (not specifically focused on 30 days), but none of these was considered truly satisfactory [13,14,15]. The fact that many patients with poor prognosis were not identified by any model was considered a major challenge. Ideally, a model would identify all or almost all patients with very short survival, and simultaneously, patients predicted to die early would not survive long enough to benefit from active treatment. In other words, both over- and undertreatment should be avoided, because shortening survival by withholding treatment would be a serious threat too.

These reflections are also applicable to the recently introduced LabPS score (blood test results and PS) [14], where the group with the poorest prognosis (3 or 3.5 points; maximum survival 2.1 months) was very small (4% of all patients in the study). Most patients with comparably short survival had a lower point sum. The LabPS score failed to outperform the previously proposed extracranial-graded prognostic assessment score (EC-GPA) [15]. Median survival was 0.7 months in the worst prognostic group of the latter score, with a hazard ratio for death of 44 (95% confidence interval (CI), 6–340) compared to the best group. However, many patients with short survival were not assigned to the worst group. After these previous, only partially successful studies that included all-comers, we changed our methodology, went back to the drawing board and hypothesized that in-depth evaluation of a carefully defined cohort with short survival, arbitrarily defined as a maximum of 3 months, may provide signals and insights, which facilitate the development of viable 30-day mortality models.

Patients and methods

A previously described, continuously updated quality-of-care database covering all adult patients with brain metastases at the authors’ institution, which employs electronic health records containing detailed follow-up information, was utilized [14]. The cohort was limited to patients who survived ≤ 3.0 months from commencing their first treatment (start of primary whole-brain radiotherapy (WBRT), date of radiosurgery (SRS), start of systemic treatment etc.), whether treatment was completed or not (intention-to-treat). The study did not include patients who received their second treatment, e.g. delayed salvage WBRT after previous SRS. Patients with leptomeningeal central nervous system metastases and those managed with best supportive care after diagnosis of brain metastases were not included. The database includes patients with solid tumors only, whereas those with leukemia and lymphoma are excluded. For the purpose of this exploratory, hypothesis-generating study, a cohort size of n = 100 was deemed appropriate. We felt that there was no solid fundament for statistical hypotheses or power calculations in the planning process. Starting with recent patients treated in 2021, backward inclusion of consecutive patients was employed. The target size of 100 patients was reached when including patients treated in the year 2011.

Besides established baseline parameters such as age, sex, number of brain metastases and Karnofsky PS, blood test results were included (hemoglobin, platelets, C-reactive protein, albumin, lactate dehydrogenase (LDH); the components of the LabBM score, which was assigned as originally recommended [16]). These were also employed to assign the LDH/albumin-based extracranial score (EC-S) [15]. The pattern and number of extracranial sites was registered (uncontrolled primary tumor, liver, lung, bone and other extracranial metastases; standard staging considered appropriate by the treating physicians at the time of treatment, thus subject to temporal and cancer-type-related variation [17]). Most likely, the blood test results mirror the overall burden of disease, including lesions not correctly identified on radiological examinations [16, 17]. The total number of brain metastases was derived from magnetic resonance imaging (MRI) reports. Cumulative lesion volume was not available.

Cause of death was recorded in order to account for surprising, unpredictable events such as accidents. Chi-square tests were employed to identify factors predicting 30-day mortality (30 days from SRS, first fraction of WBRT, day 1 of chemotherapy etc.). The latter were further examined in multinominal logistic regression analysis. Statistical significance was defined as p < 0.05 in two-sided tests. The methods employed by Rades et al. were utilized to calculate a point sum reflective of 30-day mortality [18, 19]. For example, a risk factor associated with 50% 30-day mortality was assigned 5 points, while 3 points were assigned for a factor associated with 30% 30-day mortality. The predictive accuracy of our model was evaluated using Harrell’s concordance index (Harell’s C). Harrell’s C shows perfect concordance if the value is 1, whilst a value of 0.5 indicates completely random concordance (an unserviceable model in other words).


The most common treatment approach was WBRT (30 Gy in 10 fractions, 64%; 20 Gy in 5 fractions, 12%). Eighteen percent of all patients failed to complete their prescribed treatment. Common tumor types included non-small cell lung cancer (NSCLC, 42%), malignant melanoma (12%) and breast cancer (11%). Detailed baseline characteristics are shown in Table 1. The 30-day mortality was 28% and an additional 39% died between 31 and 60 days.

Table 1 Patient characteristics, n = 100

The cause of death was unrelated to brain metastases in 61%. Both, extracranial metastases and uncontrolled primary tumors leading for example to hemoptysis or refractory pneumonia were among the documented causes of death. Brain metastases may have contributed to death in 32% (uncertainty because the patients died at home or in nursing homes; no firm documentation about the last days in our electronic patient records; both intra- and extracranial tumor activity was recorded before hospital care was terminated). Definitive confirmation of brain-related death was available in the remaining 7%, including one patient who died from hemorrhage. Treatment-related death (grade 5 toxicity) did not occur. Completely unexpected death was not observed, e.g. accident or suicide.

Univariate analyses (all factors included in Table 1 were tested; chi-square tests) revealed numerous risk factors for 30-day mortality, which were carried forward to confirmatory regression analysis. The predictive factors that achieved statistical significance in the logistic regression analysis are shown in Table 2. Based on these 9 factors (each assigned 3–6 points), a point sum was calculated for each patient. The point sum ranged from 0 (no risk factors for death within 30 days present) to 30. The results can be grouped into 3 or 4 risk categories, as displayed in Table 3. Because the model did not perform optimally (Harrell’s C 0.68; only 10 cases of 30-day mortality were assigned to the highest risk group; 10 of 28), we provided a complete data overview by tabulating the baseline parameters of all 28 patients who experienced 30-day mortality in Table 4. As illustrated in the table, 4 of 28 patients (14%) had less than two risk factors. Among them was a 93-year-old patient with uncontrolled lung cancer and hepatic metastases, whose early death would not be considered surprising by most oncologists. This example illustrates that combining a statistical model with oncological experience may be a reasonable approach.

Table 2 Factors predicting 30-day mortality (p < 0.05 in multinominal logistic regression analysis)
Table 3 Point sum leading to the final prediction model
Table 4 Factors indicating poor prognosis (bold text) in all 28 patients who died within 30 days. Typically, at least twofactors were present, e.g. poor performance status and numerous brain metastases. Four patients had less thantwo factors


After more than a decade of partially successful attempts by our group to develop and validate models that predict short survival after treatment of brain metastases, the present study represents a rigorous effort with modified methodology. We increased the number of evaluated variables, selected a narrowly defined cohort of patients with maximum survival of 3 months, and focused primarily on a dichotomized outcome (30-day mortality yes/no), which undoubtedly represents very short survival. We hoped that an in-depth analysis of a limited number of real-world patients treated with different standard approaches might pave the way towards clinically applicable risk stratification, provided external validation of the resulting model will be successful.

As demonstrated in the Results section, 30-day mortality is a highly multifactorial event. Patient-, intra- and extracranial disease-related risk factors were identified, e.g. KPS, number of brain metastases, pattern and extent of extracranial metastases, and blood test results. Interestingly, age was not associated with 30-day mortality, despite its well-known prognostic impact in analyses that looked at complete Kaplan–Meier curves [3, 4]. Given that the model did not identify or explain all instances of 30-day mortality, the real picture is probably even more complicated. This is also illustrated by the example of the 93-year-old patient included in Table 4. Reality might in fact be too complex to replace clinical judgement by partially helpful models. On the other hand, a large proportion of patients in the highest risk group (> 16 points) died within 30 days, and none survived for more than 2 months. Therefore, the model could be regarded as one of several components of decision making. As also evident from Table 4, no more than two of these 28 early deaths can be considered relatively unexpected. Causes such as accident, suicide or sudden cardiac death were not recorded.

It is also important to realize that 30-day mortality rarely was caused by the brain metastases themselves, although a certain number of patients had causes of death that remained difficult to assign. Only 5 of 100 patients did not harbor active extracranial disease, while more than 50% had at least 3 sites. In this context, one should note that we did not account for the number and size of organ lesions. Both, single bone metastases and widespread involvement were grouped under the same label (bone metastases present). Maybe, a more nuanced assessment would improve the predictive model. On the other hand, there is reason to believe that the LabBM score reflects the extracranial disease burden [16]. As suggested from our regression analysis, several measures of extracranial disease activity contributed relevant information.

A different group conducted a retrospective study of patients evaluated for palliative radiotherapy (different indications) from 2017 to 2019 who died within 90 days of consultation [20]. Data were collected for the TEACHH and Chow models and one point was assigned for each adverse factor. The TEACHH model included primary site of disease, PS, age, prior palliative chemotherapy courses, hospitalization within the last 3 months, and presence of hepatic metastases. The Chow model included non-breast primary, site of metastases other than bone only, and PS. A total of 505 patients with a median overall survival of 2.1 months were studied. Based on the TEACHH model, 2%, 77% and 21% were predicted to live > 1 year, > 3 months to ≤ 1 year, and ≤ 3 months, respectively. Utilizing the Chow model, 21%, 50% and 29% were expected to live 15.0, 6.5, and 2.3 months, respectively. Thus, neither model correctly predict prognosis in a patient population with a survival < 3 months.

External validation of our results in a larger study is necessary, given that some of the findings are surprising and based on small numbers. For example, breast cancer patients with hormone receptor-positive Her2-negative disease were at high risk, while those with triple negative disease were not. Accidental findings and overfitting of data are of concern as long as validation results are lacking. In addition, limitations include the single-institution design and the uncertainty about the cause of death in a proportion of patients. For validation studies, it would also be desirable to include intracranial tumor volume and additional surrogate markers of poor survival, e.g. hypercalcemia or cancer-related pericardial effusion or ascites. There are different ways of measuring radio- or chemotherapy utilization near the end of life, e.g. 30-day mortality calculated from start of treatment, 30-day mortality calculated from end of treatment, or treatment in the last 30 days of life. Regardless of this study’s limitations and the unique patient selection, the topic of active treatment in the terminal phase of cancer continues to be important for patients and providers alike [21,22,23,24].


Extracranial tumor progression was the prevailing cause of 30-day mortality and few, if any deaths could be considered relatively unexpected when looking at the complete oncological picture. We were able to develop a multifactorial prediction model. However, the model’s performance was not fully satisfactory and it is not routinely applicable at this point in time, because external validation is needed to confirm our hypothesis-generating findings.

Availability of data and materials

The dataset supporting the conclusions of this article is available at request from the corresponding author, if intended to be used for meta-analyses.



Performance status


Diagnosis-specific graded prognostic assessment


Extracranial-graded prognostic assessment score


Whole-brain radiotherapy


Stereotactic radiosurgery


Lactate dehydrogenase


Non-small cell lung cancer


  1. Nieblas-Bedolla E, Zuccato J, Kluger H, Zadeh G, Brastianos PK. Central nervous system metastases. Hematol Oncol Clin North Am. 2022;36:161–88.

    Article  PubMed  Google Scholar 

  2. Nieder C, Hintz M, Popp I, Bilger A, Grosu AL. Long-term survival results after treatment for oligometastatic brain disease. Rep Pract Oncol Radiother. 2020;25:307–11.

    Article  PubMed  PubMed Central  Google Scholar 

  3. Sperduto PW, Mesko S, Li J, Cagney D, Aizer A, Lin NU, et al. Survival in patients with brain metastases: Summary report on the updated diagnosis-specific graded prognostic assessment and definition of the eligibility quotient. J Clin Oncol. 2020;38:3773–84.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  4. Nieder C, Mehta MP, Geinitz H, Grosu AL. Prognostic and predictive factors in patients with brain metastases from solid tumors: a review of published nomograms. Crit Rev Oncol Hematol. 2018;126:13–8.

    Article  PubMed  Google Scholar 

  5. Schaule J, Kroeze SGC, Blanck O, Stera S, Kahl KH, Roeder F, et al. Predicting survival in melanoma patients treated with concurrent targeted- or immunotherapy and stereotactic radiotherapy: melanoma brain metastases prognostic score. Radiat Oncol. 2020;15:135.;15:280.

    Article  PubMed  PubMed Central  Google Scholar 

  6. Mulvenna P, Nankivell M, Barton R, Faivre-Finn C, Wilson P, McColl E, et al. Dexamethasone and supportive care with or without whole brain radiotherapy in treating patients with non-small cell lung cancer with brain metastases unsuitable for resection or stereotactic radiotherapy (QUARTZ): results from a phase 3, non-inferiority, randomised trial. Lancet. 2016;388:2004–14.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  7. Seymour EK, de Souza JA, Fendrick AM. Incorporating value-based care into oncology. Cancer J. 2020;26:311–22.

    Article  PubMed  Google Scholar 

  8. Schnog JB, Samson MJ, Gans ROB, Duits AJ. An urgent call to raise the bar in oncology. Br J Cancer. 2021;125:1477–85.

    Article  PubMed  PubMed Central  Google Scholar 

  9. Jones GS, Elimian K, Baldwin DR, Hubbard RB, McKeever TM. A systematic review of survival following anti-cancer treatment for small cell lung cancer. Lung Cancer. 2020;141:44–55.

    Article  PubMed  Google Scholar 

  10. Yoo TG, Cranshaw I, Broom R, Pandanaboyana S, Bartlett A. Systematic review of early and long-term outcome of liver resection for metastatic breast cancer: Is there a survival benefit? Breast. 2017;32:162–72.

    Article  PubMed  Google Scholar 

  11. Krishnan MS, Epstein-Peterson Z, Chen YH, Tseng YD, Wright AA, Temel JS, et al. Predicting life expectancy in patients with metastatic cancer receiving palliative radiotherapy: the TEACHH model. Cancer. 2014;120:134–41.

    Article  PubMed  Google Scholar 

  12. Nieder C, Angelo K, Dalhaug A, Pawinski A, Haukland E, Norum J. Palliative radiotherapy during the last month of life: predictability for referring physicians and radiation oncologists. Oncol Lett. 2015;10:3043–9.

    Article  PubMed  PubMed Central  Google Scholar 

  13. Nieder C, Marienhagen K, Thamm R, Astner ST, Molls M, Norum J. Prediction of very short survival in patients with brain metastases from breast cancer. Clin Oncol (R Coll Radiol). 2008;20:337–9.

    Article  CAS  Google Scholar 

  14. Nieder C, Yobuta R, Mannsåker B. Expansion of the LabBM score: is the LabPS the best tool predicting survival in patients with brain metastases? Am J Clin Oncol. 2021;44:53–7.

    Article  PubMed  Google Scholar 

  15. Nieder C, Hess S, Lewitzki V. External validation of a prognostic score for patients with brain metastases: extended diagnosis-specific graded prognostic assessment. Oncol Res Treat. 2020;43:221–7.

    Article  CAS  PubMed  Google Scholar 

  16. Berghoff AS, Wolpert F, Holland-Letz T, Koller R, Widhalm G, Gatterbauer B, et al. Combining standard clinical blood values for improving survival prediction in patients with newly diagnosed brain metastases–development and validation of the LabBM score. Neuro Oncol. 2017;19:1255–62.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  17. Nieder C, Mehta MP, Guckenberger M, Gaspar LE, Rusthoven CG, Sahgal A, et al. Assessment of extracranial metastatic disease in patients with brain metastases: how much effort is needed in the context of evolving survival prediction models? Radiother Oncol. 2021;159:17–20.

    Article  PubMed  Google Scholar 

  18. Rades D, Dunst J, Schild SE. A new scoring system to predicting the survival of patients treated with whole-brain radiotherapy for brain metastases. Strahlenther Onkol. 2008;184:251–5.

    Article  PubMed  Google Scholar 

  19. Rades D, Dziggel L, Haatanen T, Veninga T, Lohynska R, Dunst J, et al. Scoring systems to estimate intracerebral control and survival rates of patients irradiated for brain metastases. Int J Radiat Oncol Biol Phys. 2011;80:1122–7.

    Article  PubMed  Google Scholar 

  20. Mojica-Márquez AE, Rodríguez-López JL, Patel AK, Ling DC, Rajagopalan MS, Beriwal S. External validation of life expectancy prognostic models in patients evaluated for palliative radiotherapy at the end-of-life. Cancer Med. 2020;9:5781–7.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  21. Ryoo JJ, Batech M, Zheng C, Kim RW, Gould MK, Kagan AR, et al. Radiotherapy for brain metastases near the end of life in an integrated health care system. Ann Palliat Med. 2017;6:S28-38.

    Article  PubMed  Google Scholar 

  22. Park KR, Lee CG, Tseng YD, Liao JJ, Reddy S, Bruera E, et al. Palliative radiation therapy in the last 30 days of life: a systematic review. Radiother Oncol. 2017;125:193–9.

    Article  PubMed  Google Scholar 

  23. Pitson G, Matheson L, Garrard B, Eastman P, Rogers M. Population-based analysis of radiotherapy and chemotherapy treatment in the last month of life within regional Australia. Intern Med J. 2020;50:596–602.

    Article  PubMed  Google Scholar 

  24. Vázquez M, Altabas M, Moreno DC, Geng AA, Pérez-Hoyos S, Giralt J. 30-day mortality following palliative radiotherapy. Front Oncol. 2021;11:668481.

    Article  PubMed  PubMed Central  Google Scholar 

Download references


Not applicable.


Open Access funding provided by UiT The Arctic University of Norway The publication charges for this article have been funded by a grant from the publication fund of UiT The Arctic University of Norway.

Author information

Authors and Affiliations



CN, LS, SGA and BM analyzed and interpreted the patient data. CN and ECH drafted the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Carsten Nieder.

Ethics declarations

Ethics approval and consent to participate

As a retrospective quality of care analysis, no approval from the Regional Committee for Medical and Health Research Ethics (REK Nord) was necessary. This research project was carried out according to our institutions’ guidelines and with permission to access the patients' data.

Consent for publication

Not applicable.

Competing interests

SGA has received lecture fees from Bristol Myer Squibb and Astra Zeneca. All other authors declare no conflict of interest.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Nieder, C., Stanisavljevic, L., Aanes, S.G. et al. 30-day mortality in patients treated for brain metastases: extracranial causes dominate. Radiat Oncol 17, 92 (2022).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: