Do Patients with a Poor Karnofsky Performance Status Scale Pro t from Tumour Volume Reduction?

Melanie Barz (  melanie.barz@tum.de ) Klinikum rechts der Isar der Technischen Universitat Munchen Neuro-Kopf-Zentrum https://orcid.org/0000-0001-8734-9309 Julia Gerhardt Helios Klinikum Berlin, Department of Neurosurgery Stefanie Bette Universitätsklinikum Augsburg, Department of neuroradiology A. Kaywan Aftahy Technische Universität München, Department of neurosurgery Thomas Huber Universitätsklinikum Mannheim: Universitatsklinikum Mannheim Stephanie E. Combs Technische Universität München: Technische Universitat Munchen Yu-Mi Ryang HELIOS Klinikum Berlin-Buch Benedikt Wiestler Technische Universität München: Technische Universitat Munchen Marco Skardelly Universitätsklinikum Tübingen: Universitatsklinikum Tubingen Irina Gepfner-Tuma Technische Universität München: Technische Universitat Munchen Felix Behling Technische Universität München: Technische Universitat Munchen Friederike Schmidt-Graf Technische Universität München: Technische Universitat Munchen Bernhard Meyer Technische Universität München: Technische Universitat Munchen Jens Gempt Technische Universität München: Technische Universitat Munchen


Introduction
In 1949, Karnofsky and Burchena described their instrument, the Karnofsky Performance Scale (KPS) score, as a numerical scale for quantifying patients' status in relation to the degree of their independence in daily activities and self-care. Originally, it was used for patients with systemic malignancies and divided them according to their level of activity and medical requirements. Patients are scored into 11 categories from 0 to 100, where, for example, a KPSS of 70% means the patient is able to care for himself but is unable to carry out daily activities [13]. After it had been proven successful in patients with systemic cancer, more and more research groups started to evaluate the KPS score for brain cancer [24,22,14].
Previously published studies could show a signi cant correlation between the preoperative KPS score and the outcome after glioma surgery [4,15]. In most studies, only patients suffering from a glioblastoma with a KPSS of ≥ 70% were included [19,6]. For example, those studies analysed prognostic factors such as tumour size, GRT and adjuvant therapy modalities postoperatively. However, in our clinical daily work, patients with a noticeably lower KPSS are represented as well. It should be noticed that this can be due to clinic symptomology as seizures, acute mental status changes or focal neurologic de cits caused by tumour size and/or location itself. Therefore, the following study intends to show whether it is worthwhile for patients with a KPSS 60% or below to achieve tumour volume reduction.

Patients And Methods
This retrospective, non-interventional bicentric study was approved by the local medical ethics committee  and is in accordance with the ethical standards of the 1964 Declaration of Helsinki and its later amendments [9].

Patient Population
We retrospectively assessed 968 patients with a histologically con rmed glioblastoma WHO IV with a preoperative Karnofsky Performance Status Scale (KPSS) of ≤ 60%, who were treated surgically between September 2006 and March 2016 in two neurosurgical departments. According to interdisciplinary neurooncological consensus, patients were assigned to surgery with the intent of complete resection or to biopsy to con rm the histopathological diagnosis. We retrospectively reviewed pre-and postoperative KPSS, date of initial tumour diagnosis, date of death/last contact, age, sex, adjuvant treatment and histopathological ndings from the patients' medical charts. Also, we performed histopathological analysis according to the WHO criteria of 2016 [17] and quantitatively assessed methylation of the O6methylguanin-DNA-methyltransferase (MGMT) promoter status. We assessed KPSS with regards to hospital admission and ve days after surgery. Then, we calculated the overall survival (OS) from the date of surgery until the date of death or censored for the date of the last patient contact. Only patients with complete magnetic resonance imaging data were included to calculate pre-and postoperative contrast-enhancing tumour volumes. Patients with recurrent tumour or incomplete data were excluded ( Fig. 1).

Imaging
All patients received preoperative and early postoperative MRI (within 72 hours after surgery). In centre A, we performed imaging using three different 3 T MRI scanners: Philips Achieva; Philips Ingenia (Philips Medical Systems, The Netherlands B.V.); and Siemens Verio (Siemens Healthcare, Erlangen, Germany).
Images included T1w sequences with and without contrast agent, FLAIR (Fluid attenuated inversion recovery) sequences, T2 gradient echo sequences, diffusion-weighted imaging or diffusion-tensor imaging, whereas we calculated isotropic diffusion-weighted images and apparent diffusion coe cient (ADC) maps automatically. Tumour volumes of the contrast-enhancing tumour on pre-and early postoperative MR images using iPlannet® Cranial 3.0.1 were manually segmented by two neurosurgeons (5 and 10 years of experience) and two neuroradiologists (3 years and 6 years of experience).
In centre B, we conducted MR imaging with a 3.0 T MRI scanner (Biograph mMR, Siemens Healthcare, Erlangen, Germany). One neurosurgeon (14 years of experience) and one medical student assessed the volumes of the contrast-enhancing tumour through manual segmentation via iPlannet® Cranial 3.0.1 (iPlannet® 3.0 cranial planning software, Brainlab AG, Munich, Germany). The postoperative tumour volumes of patients who underwent biopsies were considered identical to the preoperative tumour volumes.

Statistical Evaluation
We conducted our data analysis using IBM SPSS Statistics Version 24.0 (SPSS Inc., IBM Corp., Armonk, NY, USA). In the descriptive data analysis, we show non-normally distributed data as median and interquartile range (IR), normally distributed variables as mean and standard deviation.
We compared the OS distributions using the Kaplan-Meier estimates (log-rank) and a Cox regression model for multivariate survival analysis. We considered differences with an error probability of less than 0.05 to be statistically signi cant.

Discussion
In this cohort of GBM patients with a preoperative KPS ≤ 60%, postoperative tumour volume, age at the time of surgery and MGMT-methylation status were signi cant predictors of OS in the multivariate analysis. In contrast, preoperative tumour volume and KPSS had no signi cant impact on OS.
Nevertheless, as already understood from other studies, we could also show that extent of resection is an important factor in OS in patients with glioblastoma [16,10,3,2].
In general, patients with poor preoperative KPSS usually do not receive aggressive surgical therapy. Therefore, data on these patients are very limited. In our cohort, 56/123 (45.5%) showed an improved or unchanged postoperative KPSS with a median of 50%. Adjuvant treatment such as radiation therapy or chemotherapy is usually only offered to patients with a KPSS ≥ 70% [20,12]. Consequently, these patients are usually considered ineligible for adjuvant oncological treatment even after tumour resection. Malakhov et al. could show that 51.2% of the patients presenting with KPSS < 60 and receiving chemoradiation had improved survival compared to RT alone [18]. However, the majority of our patient cohort (77.6%) who underwent surgical resection received adjuvant therapy. Considering the early postoperative assessment of KPSS in this study, secondary improvement is to be expected. Patients undergoing a biopsy were older and had smaller preoperative tumour volumes than patients, who were selected for surgical tumour resection. Only 16.7% of the patients who received a biopsy underwent adjuvant treatment regimes.
Reduced preoperative KPSS is an important prognostic factor in patients with glioblastoma [21,23]. Age, comorbidities and neurological de cits have an impact on KPSS and, in conclusion, on OS [1,5,23].
Postoperative deterioration of the performance status scale is usually multifactorial, with the reasons being edema, haemorrhage, postoperative delirium, ischemic events or direct surgical lesions of eloquent brain structures [8].
In our opinion, the KPSS does not offer su cient information about quality of life and therefore should not be overrated concerning the selection of patients undergoing surgery. For example, patients with preoperative neurological de cits such as hemiparesis due to surrounding edema might have a KPSS of 60% or below and might therefore not be selected for surgical therapy. However, as we know today, the surrounding edema will disappear a few days after surgery, and the patients are able to recover for adjuvant treatment. The KPSS should therefore be considered with care.
The decision for or against aggressive surgical therapy should be made individually by experienced neurosurgeons within the framework of an interdisciplinary neuro-oncology board.

Limitations of the Study
This study has limitations. First, the retrospective non-randomized design is the main limitation. Second, molecular status was not available for all patients in our cohort study, as the MGMT-methylation status of patients with glioblastoma is known to be one of the strongest predictors concerning survival prognosis [7,11].

Conclusion
Even glioblastoma patients with a poor preoperative KPSS seem to pro t from low postoperative residual tumour volumes. Age at the time of surgery and MGMT-methylation status had a signi cant in uence on OS in our series. We therefore suggest considering surgical resection even in patients with a KPSS of ≤ 60% after careful selection based on an interdisciplinary neuro-oncological board decision and counselling of patients and their relatives. In addition, BM works as a consultant for Medtronic, Spineart, Icotec, Relievant and Depuy/Synthes. In these rms, BM acts as a member of the advisory board. Furthermore, BM reports a nancial relationship with Medtronic, Ulrich Medical, Brainlab, Spineart, Icotec, Relievant and Depuy/Synthes. He received personal fees and research grants for clinical studies from Medtronic, Ulrich Medical, Brainlab, Icotec and Relievant. All this occurred independently of the submitted work. BM holds the royalties/patent for Spineart.

Declarations
All named potential con icts of interest are unrelated to this study.

Ethical Approval
This retrospective, non-interventional bicentric study was approved by the local medical ethics committee (5625-12) and is in accordance with the ethical standards of the 1964 Declaration of Helsinki and its later amendments [9].

Informed Consent
All patients sign a generally valid declaration of consent for participation in retrospective studies upon admission.