A pilot study to quantify and describe personal social networks in patients with glioblastoma

Abstract

Background

Patients with glioblastoma experience high physical and psychosocial symptom burden. Poor social relationships have been shown to increase the risk of neurologic illnesses and decline, and conversely, strong personal social networks (PSN) have been shown to reduce the risk of mortality and improve quality of life. The aim of this pilot study is to determine the feasibility of measuring PSN in glioblastoma patients.

Methods

We recruited 25 adult glioblastoma patients between the initial diagnosis and the first cycle of adjuvant chemotherapy from March to September 2023 in the outpatient neuro-oncology clinic and adapted PERSNET, a quantitative PSN assessment tool, to this population. We collected demographics, tumor measures, treatment courses, and the European Organization for Research and Treatment of Cancer QOL for Patients with Brain Tumors (EORTC QLQ-BN20) and conducted qualitative interviews.

Results

The average age was 64.5 years old, 56% were female, and 84% had a Karnofsky Performance Status of 70 or higher. Patients had large network sizes (total size of patient’s PSN; mean = 8.8). Participants stressed the importance of social support and how different people filled different roles in their oncology care. Behavioral and/or cognitive changes resulted in delayed presentation, and children, especially daughters, were important in advocating for presentation to the hospital.

Conclusion

This is the first proof-of-concept study showing that PSN can be measured in patients with glioblastoma. Qualitative data showed that patients felt social support was very important, and different people in their networks addressed different domains of care: physical, emotional, and logistical.

Graphical Abstract

Graphical Abstract.

Key Points.

• Personal social networks can be quantified in glioblastoma patients.

• Different people filled different support roles for the patient.

• Family members, especially daughters, were important in earlier presentation.

Importance of the Study.

This pilot study is the first to study personal social networks (PSN) in glioblastoma patients. Our qualitative data showed that different parts of a patient’s PSN filled different needs in the emotional and physical care of the individual patient and that patients with cognitive changes had delayed presentation to the hospital. PSN can be a tool to better understand how social supports may evolve over the course of an illness where neurologic deficits may affect relationships and interactions. Future studies of PSN in glioblastoma patients should focus on measuring how PSNs change throughout the course of the disease and how changes in PSNs affect the quality of life for patients. Understanding the PSNs of a patient with glioblastoma may help determine when and how to best support them. As was shown in other studies, understanding a patient’s social environment could influence behavioral and social interventions aimed at improving health outcomes in glioblastoma patients.

Patients with glioblastoma experience high physical^1–3 and psychosocial symptom burden⁴ as well as a high fatality rate.⁵ Furthermore, progressive neurologic deficits in language, strength, and gait restrict a patient’s ability to interact with others, potentially leading to increased social isolation. Research has shown that patients’ limited ability to attend social events and participate in social interactions leads to feelings of social isolation,⁶ which is compounded as physical and cognitive symptoms worsen over the course of a patient’s illness.⁷ Personality changes are often distressing for caregivers of patients with glioblastoma and further exacerbate social isolation.⁷

Research has increasingly established the importance of one’s social environment in determining health outcomes.⁸ For example, social isolation is a known determinant of poor health⁹ and socioeconomic and behavioral factors in chronic disease are major contributors to health outcomes.¹⁰ In patients with neurological diseases, decreases in function, ability to communicate and think, and mobility all contribute to a patient’s isolation.⁹ Additionally, poor social relationships have been shown to increase the risk of neurological diseases, with a 32% increased risk of stroke and 50% to 100% increased risk for developing dementia.^11,12 On the other hand, studies have shown that strong personal social networks (PSNs) reduce the risk of mortality and improve quality of life (QOL), particularly in high-stress situations,^13,14 and increased PSNs may protect against cognitive decline.¹⁵ Social supports and structures appear to be especially important for the well-being of patients with glioblastoma, as neurologic decline may affect PSNs. The predicted progressive nature of this illness also places a heavy burden on both patients and caregivers.^4,7 In fact, increased caregiver mastery has been shown to even be associated with increased survival in glioblastoma patients.^16,17 It is, therefore, crucial to study reliable, quantitative methods of assessing social support structures and their impact on patients’ experiences of living with glioblastoma.

PERSNET is a standardized survey that quantitatively measures the PSN of patients with neurological illnesses.⁸ In their investigation of social network influences on patient outcomes, Dhand and colleagues⁸ found that small, close-knit PSNs of highly familiar contacts led to constricted information flow resulting in delayed hospital arrival after acute stroke.¹⁸ This methodology has since been replicated in other contexts, including in Down’s Syndrome,¹⁹ tele-rehabilitation,²⁰ and head and neck cancers,²¹ demonstrating the feasibility of quantitatively assessing the PSN in these patient populations using the same methodology.²² A participant’s PSN may be visualized as a sociogram (Supplementary Appendix 1), serving as a basis for further quantifying PSN characteristics such as network size, composition, and the degree of interconnectedness (ties) among members of the patient’s network.

In PERSNET, the primary individual to whom the network belongs to, in our case, the patient, is known as the ego.⁹ Individuals in the network, excluding the ego, are referred to as the alters.⁹ In our study, the alters were often friends or family. The network size is the total amount of alters in the network.⁹ Density is a measure of connectivity of individuals in the network based upon the quantity of ties between the alters.^9,19 Constraint is a measurement of how tightly connected the members of the network are to the network in general or the extent to which the ego’s connections are to individuals who are also connected to one another.^9,19,23 Effective size is the number of unique informational “groups” that exist within the network or the inverse of constraint.^9,19 For example, if a patient has connections with several distinctive non-overlapping social groups, then their effective network size will be greater and their network constraint score will be lower compared to a patient with the same number of connections, whose ties are connected to one another, like a large family.^9,19 Maximum degree is the highest number of ties by a network member and the mean degree is the average number of ties by a network member.^9,19

This current study was a pilot, proof-of-concept study where we studied the PSNs of adult glioblastoma patients. The primary outcome of this study was to determine the feasibility of measuring PSNs in glioblastoma patients. We achieved this by quantifying patients’ PSNs with PERSNET and various demographic variables, clinical outcomes, tumor characteristics, functional status, and QOL measures as well as conducting qualitative interviews with participants. Future studies will be aimed at investigating the relationship between these variables with sufficient statistical power in the next phase of this study.

Methods

Patients

This study received Institutional Review Board approval under DFCI IRB #21-359. We recruited patients from the Dana-Farber Cancer Institute outpatient neuro-oncology clinic from March 2023 to September 2023. Of the 29 patients who were approached and consented, 25 patients completed the study. We approached newly diagnosed patients with glioblastoma after their clinic visit for informed consent, followed by a one-time phone or video structured interview. This interview was scheduled at the participant’s earliest convenience. All participants completed the interview prior to their first adjuvant cycle of chemotherapy, with the large majority completing the interview within a week of consent (88%). The interview included the PSN survey (PERSNET) with the patient and, if the patient desired, their primary caregiver. Primary caregivers were often family or friends and were allowed to be included in the interview, clarifying specific information about the PSN that the participant may have difficulty recalling. In addition, we used the electronic health record to collect demographics, socioeconomic status, Karnofsky Performance Scale (KPS) at the time of diagnosis, specific treatments for glioblastoma, time to presentation (by history), and size of the tumor at presentation (measured by a neuro-oncologist and radiologist) as a proxy for time to arrival.

Patients who were 18 years old or older at the time of initial glioblastoma diagnosis were able to participate. Patients had to have a histopathological confirmation of a glioblastoma diagnosis at the time of informed consent and had to have the ability to understand and sign a written informed consent form.

PERSNET

The PSN survey (PERSNET) is a mixed-methods survey that was previously created, validated, and implemented by Dhand et al.¹⁸ For our patient population, we conducted PERSNET at a single time point between initial diagnosis and the patient’s first cycle of adjuvant chemotherapy. PSNs may be understood in terms of structure (size and shape of the social connections) and composition (characteristics of people around the index patient). A detailed explanation of the components of the PERSNET has already been published.^9,18 Supplementary Appendix 2 includes the equations used to calculate each network characteristic.

EORTC QLQ-BN20

Immediately after completing the PERSNET with participants, we collected additional data using a validated QOL measure, the European Organization for Research and Treatment of Cancer (EORTC) survey for patients with brain tumors (EORTC QLQ-BN20),.²⁴ The EORTC QLQ-BN20 is a validated health-related QOL questionnaire specific to patients with brain neoplasms and validated in multiple languages.²⁴ We conducted Pearson’s correlation test between each of the EORTC QLQ-BN20 categories with network characteristics using IBM SPSS statistics software. The level of significance for 2-tailed tests was set at P < .05, corresponding to a 95% confidence interval. We did this to assess the strength and direction of correlations to see if there were any correlations that should be investigated in future studies and to inform the methods of those studies.

Qualitative Analysis

The PERSNET questionnaire included 2 qualitative questions: (1) How do the people you mentioned influence your quality of life after diagnosis, and (2) What were the biggest factors that affected how quickly you sought out medical care? One of the co-authors (CM) asked the question, and patients and/or caregivers opted to type their responses in the PERSNET survey or have the researcher paraphrase their responses in the PERSNET survey.

The responses were analyzed using thematic analysis to identify and define the emerging common themes.²⁵ JR and ZT both independently read all the responses, and both independently coded all responses from the data. Following coding, we identified, reviewed, and refined the main themes and subthemes throughout the analysis, and the results were triangulated by comparing and discussing emerging themes.

Results

We enrolled 29 patients, of which 25 patients completed the study. The average age of participants was 64.5 years old (range 20.5–86.1), with a median age of 63.6 (IQR 60.0–73.4). About half (14; 56%) of participants were female, the large majority (24; 96%) were white, and one-fifth (5; 20%) identified as Hispanic. The highest degree earned varied from about a third having received a high-school diploma or GED (8; 32%) and a quarter having received higher than a bachelor’s (6; 24%) (Table 1).

Table 1.

Demographic, Tumor, and Treatment Characteristics of the Patient Population.

Background characteristics	Result (n = 25) % unless otherwise indicated)
Age	Mean 64.5 years old Range 20.5–86.1 Median: 63.6 IQR: 60.0–73.4
Female	14 (56.0%)
Highest Degree Earned
High School Diploma/GED	8 (32%)
Associate’s	3 (12%)
Bachelor’s	7 (28%)
Doctorate/Professional	6 (24%)
None of the above	1 (4%)
Annual Income Earned
<$15 000	0 (0%)
$15 000–$24 999	2 (8%)
$25 000–$49 999	3 (12%)
$50 000–$74 999	3 (12%)
$75 000–$99 999	3 (12%)
>$100 000	14 (56%)
Race
White	24 (96%)
Other	1 (4%)
Hispanic	5 (20%)
Marital Status
Never Married	1 (4%)
Married	14 (56%)
Domestic Partnership	1 (4%)
Divorced/Separated	7 (28%)
Widowed	2 (8%)
Live alone	4 (16%)
Tumor laterality
Left	13 (52%)
Right	10 (40%)
Bilateral	1 (4%)
Multifocal	1 (4%)
Tumor lobe
Frontal	11 (44%)
Parietal	10 (40%)
Occipital	0 (0%)
Temporal	4 (16%)
Multifocal	2 (8%)
MGMT- methylated (n = 24)	13 (54.2%)
Radiation
Standard (6 weeks)	13 (59.1%)
Hypofractionated	9 (40.9%)
Other	3 (12%)
Concurrent temozolomide	20 (83.3%)
Upfront bevacizumab	1 (4%)
Size of initial tumor in cm3 (mean [SD])	71.65 [72.31]
Time between symptoms to pathologic diagnosis of glioblastoma in days (mean [SD]; median)	47.76 [51.98]; 22
Karnofsky performance status at the first neuro-oncology visit
<30	0 (0%)
40–60	4 (16%)
70–90	18 (72%)
100	3 (12%)

More than half of the participants (13; 59.1%) received the standard course of radiation (6 weeks), while 40.9% (n = 9) received a hypo-fractionated course of radiation. Three patients (12%) either decided to forgo radiation for hospice or went on straight to only chemotherapy. One patient received upfront bevacizumab at the time of initiation of radiation. About a quarter (5; 22.7%) of patients enrolled in a clinical trial (Table 1).

The most prominent symptoms prior to hospital presentation were headache (36%), fatigue (32%), and confusion (28%), while the most prominent symptoms at presentation were fatigue (59%), headache (32%), and weakness (27%) (Table 2).

Table 2.

Symptoms and EORTC QLQ-BN20 Categories of the Patient Population.

Symptoms	Prior to hospitalization (n = 25, %)/At presentation (n = 22, %)
Headache	9 (36)/7 (31.8)
Fatigue	8 (32)/13 (59.1)
Confusion	7 (28)/3 (13.6)
Seizure	6 (24)/4 (18.2)
Behavioral/Personality changes	6 (24)/0 (0)
Weakness	4 (16)/6 (27.3)
Unsteady gait	4 (16)/4 (18.2)
Visual changes	3 (12)/3 (13.6)
Memory loss	3 (12)/3 (13.6)
Nausea/vomiting	1 (4)/3 (18.2)
Tremors/shaking	1 (4)/0 (0)
Numbness	0 (0)/3 (13.6)
Dizziness	0 (0)/2 (9.1)
Other	9 (36)/8 (36.4)
EORTC QLQ-BN20 Categories	Median (IQR)
Future uncertainty (BNFU)	33.33 (25–66.67)
Visual disorder (BNVD)	0 (0–22.22)
Motor dysfunction (BNMD)	22.22 (0–55.56)
Communication deficit (BNCD)	11.11 (0–33.33)
Headaches (BNHA)	0 (0–33.33)
Seizures (BNSE)	0 (0–0)
Drowsiness (BNDR)	33.33 (33.33–66.67)
Itchy skin (BNIS)	0 (0–33.33)
Hair loss (BNHL)	0 (0–0)
Weakness of leg (BNWL)	0 (0–33.33)
Bladder control (BNBC)	0 (0–0)

Personal Network Characteristics

We administered the PERSNET to all 25 participants. More than half of the participants were married or in domestic partnerships (15; 58%), and 16% (4 participants) reported living alone. There was also strong connectivity of individuals within the network, measured through density (0.83) and constraint (50.10) (Table 3). While patients had, on average, 7.16 alters, the average effective network size was 3.29, suggesting that most patients listed ties who also knew each other (ie, families or circles of friends). We were also interested in any differences in network demographic characteristics such as sex. We did a sub-analysis comparing males and females along the same PSN characteristics and found minor differences in trends in the medians of all network characteristics, though differences were not statistically significant (Supplementary Appendix 3).

Table 3.

Net Personal Social Network Characteristics

	Mean (SD)	Median (IQR)	Min-Max
Network size (nodes)	8.80 (4.37)	8 (5–11)	2–19
Edges	26.56 (13.90)	27 (15–36)	3–55
Density	0.83 (0.22)	1 (0.7–1)	0.33-1
Constraint	50.10 (17.97)	44.44 (37.05–57.77)	31.59–112.50
Effective size	3.29 (1.78)	3 (2–3.6)	1–6.78
Maximum degree	5.84 (2.12)	6 (4–7)	1–9
Mean degree	3.76 (2.79)	4 (1–6)	0–9

Figure 1 shows the sociograms of each individual patient in our cohort. Line color indicates the strength of the relationship, with red lines indicating strong ties and blue lines indicating weak ties. As Figure 1 shows, there is a wide variety in PSN sizes and connectivity.

Figure 1.

Sociograms of the personal social networks of glioblastoma patients (n = 25). In each network, the central dot represents the ego with glioblastoma, outer dots represent other social connections or the alters, dashed lines represent weak social ties between two persons, and solid lines represent strong social ties between two persons.

We chose 2 participants on the lower and upper ends of the network size and constraint distributions and described them, in depth, to provide case models of 2 patients with varying sociograms (Figure 2). Specifically, we were interested in the differences between participants with large versus small network sizes to see on an individual level how network size affected these individuals. Both participants had the same density of 1, suggesting that the PSNs were equally well connected but had varying constraint, meaning that people in the network were on varying ends of how connected they were to one another.

Figure 2.

Demographics, personal social network characteristics, and qualitative PERSNET data of participants 11 and 13.

Interestingly, though Participant 13 had a lower KPS of 70, more symptoms, and a larger tumor (121,824 mm³) compared to Participant 11 (KPS of 90, tumor size of 24,800 mm³), Participant 13 reported better QOL in the qualitative questions as well as EORTC measures regarding future uncertainty and thinking positively about the future (Figure 2). This reflects that PSNs may have a strong weight in the QOL of a patient, especially as it has to do with worries and/or anxieties about the future.

EORTC QLQ-BN20

When the EORTC QLQ-BN20 questions (Supplementary Appendix 4) were categorized into common themes per EORTC analysis guidelines, on average, patients scored higher on future uncertainty as a source of distress, and with regards to symptoms, drowsiness as the symptom with the highest burden, followed by motor dysfunction (Table 2). Overall, the range of experience of symptoms was very wide, with some patients relatively asymptomatic at diagnosis compared to others who were already fairly symptomatic at the point of diagnosis.

In an exploratory Pearson’s correlation test between EORTC QLQ-BN20 categories and PERSNET characteristics (Supplementary Appendix 5), we found a positive correlation between hair loss and constraint (r = 0.506, P = .010) as well as a negative correlation of hair loss with edges (r = −0.442, P = .027), maximum degree (r = −0.435, P = .030), and mean degree (r = −0.397, P = .049). Motor dysfunction was also positively correlated with mean degree (r = 0.435, P = .030).

Qualitative Analysis

Question 1: How do the people you mentioned influence your quality of life after diagnosis?

Several themes emerged for question 1: (1) different people met different needs, (2) social support is very important, and (3) relationships may cause distress.

• (1) Different people met different needs.

Participants noted that different people in their personal social network met different needs. For example, some people provided physical support, while others provided emotional support, advice, or sharing time together over meals. Participants noted particular people by name and what type of support they offered.

“[Name] and [name] have influenced it greatly […] [Name] and [name] and [name] aren’t good at providing emotional support but are very available for any physical needs I need and camaraderie. [Name] has also improved my quality of life in many ways.” – Participant #5

• (2) Social support is very important.

Most participants noted that social support, in whatever form, was very important, and that they found it helpful. Adjectives used to relay the gratefulness for support included “tremendously,” “helpful,” “very, very supportive.”

“Strength and hope knowing they would be there to do anything I needed.” – Participant #6

• (3) Relationships may cause distress.

In a few cases, participants also brought up where current and prior relationships caused additional distress. One noted grief at certain family members not offering enough support or going through their own grief and/or processing of the disease.

“[Name] has influenced it well but also gives me stress because of her anxiety and grief regarding my diagnosis.” – Participant #5

“[I feel I have] no support from [my] husband.” -Participant #11.

Question 2: What were the biggest factors that affected how quickly you sought out medical care?

For question 2, the following additional themes emerged: (1) a range of presenting symptoms, (2) family members were first to notice changes.

• (1) There is a range of presenting symptoms.

Participants noted a range of presenting symptoms, including what they interpreted as seizure, mini-stroke, behavioral and/or cognitive changes, forgetfulness, changes in speech, heavy vomiting, headache, changes in controlling a body part, and fall. Out of these symptoms, behavioral, cognitive, and/or memory changes were mentioned most frequently (8 times) followed by seizure and speech changes (4 times each). In all cases, they were described as compared to the patient’s baseline. Changes in memory and/or behavioral changes often led to delay in presentation and subsequent diagnosis whereas more acute symptoms such as seizure, stroke, vomiting, fall, led to a more expedited presentation to the hospital.

“And then he had some episodes of confusion, he would go for walks and lose track of where he was, walk down someone’s driveway and not realize. Chalked it up to being distracted / in his own head when walking around. Then he was in house and walked into another room and did not know where he was.” – Participant #24

• (2) Family members were first to notice changes.

In most cases, family members were first to notice changes, and children, especially daughters, were most frequently mentioned as being the first to notice changes and persuade the patient to present to the hospital.

“[I] explained symptoms over phone to [my] daughter. Daughter urged [me] to call [my] PCP who advised visit to urgent care, and was then transferred to ER.” – Participant #11

Discussion

This proof-of-concept study demonstrated that the PSNs of patients with glioblastoma can be described quantitatively. Quantitative analyses revealed that glioblastoma patients in our study have large network sizes at initial diagnosis. The qualitative analysis showed the importance of larger networks in providing various forms of social support (eg, emotional, physical, transport) to the patient. Another finding of the qualitative data was that while family members were the first to notice the change, patients were quicker to present to the hospital when the symptoms were acute (eg, seizure, stroke, vomiting, fall) as opposed to changes in memory and/or behavioral changes.

These findings replicate and build on previous studies^8,9,18,19 providing evidence for the possibility of measuring social support among neurologic patient populations. Additionally, the influence of social environments on health outcomes⁸ suggests that focusing on PSNs could be a valuable approach to improving glioblastoma patient outcomes. Furthermore, supportive care trials often aim to improve support structures for patients but rarely utilize quantitative measures to determine whether a certain behavioral intervention can change or expand those supports. PSNs show promise as a standardized measure in supportive care trials, providing a better understanding of how various interventions may affect psychological well-being or QOL outcomes.²² In cancer care, particularly in brain cancer, neurological deterioration often necessitates large social networks to support the patient through emotional comfort and practical support, such as facilitating treatment and offering respite for caregivers.^4,7 A validated quantitative measure in this population would be instrumental in studies on optimizing patient support during treatment, facilitating meaningful relationships for demoralized individuals, and providing caregivers with additional support structures. Furthermore, the act of describing their PSN may be an intervention in and of itself as reflecting on the people supporting them may help bring about a sense of gratitude and reduce negative thoughts and emotions.

In our study, the PSNs of patients with glioblastoma had large network sizes (median 8.8, IQR 5.0–11.0) compared to stroke (slow arrivers network size median 5.0, IQR 4.0–8.0; fast arrivers network size median 8.0, IQR 6.0–11.0)¹⁸ and Down syndrome (network size median 7.5; IQR 5.25–11.0).¹⁹ This may be due to an initial increase in perceived personal network size after diagnosis because the support network may be easily accessed and activated due to public perceptions around the diagnosis of cancer.²⁶ However, future research should investigate how the network size changes over time and whether at recurrence, when physical and communication symptoms are more challenging, the network characteristics remain stable. Additionally, it would be helpful to see whether network size is the most important variable for QOL or whether other network characteristics, such as how the network is connected to one another (such as constraint) may be more important in predicting QOL. Lastly, it would be important to see whether network characteristics may be related to cancer outcomes such as overall survival or progression-free survival.

The qualitative analysis reflected that a larger PSN may have been more beneficial to glioblastoma patients as patients reported that various people in their network fulfilled different types of social support, ranging from emotional versus physical support, having meals together, and providing advice. Furthermore, patients overwhelmingly affirmed that PSN, in whatever form, were important and helpful in dealing with their illness, even though relationships can, at times, cause stress. This further highlights the importance of studying PSNs, and how they may affect a patient’s overall well-being and QOL.

Our case studies of 2 patients, representing the upper and lower ends of network size also provided additional interesting insights. Participant 13, though having more physical symptoms than participant 11, felt more hopeful about the future, more secure, and reported better QOL, which may be from their larger PSN. Hypothetically a larger network can make an individual feel more secure and less distressed about their life with glioblastoma. Should a larger PSN have these positive effects on patients, clinicians could focus more on social support systems for patients to improve health outcomes. Further studies should test the hypothesis that social support may be a strong determinant of health outcomes and may even offset the poorer QOL attributed to worse physical symptoms.

Though both participants 11 and 13 reported that their PSNs were supportive, participant 11 reported extreme unhappiness when not visited by the members of their network. This could suggest that smaller networks may struggle to meet the various needs of an individual patient, as shown by the qualitative data indicating that different people or groups fulfill different emotional or physical needs. In other words, if PSNs provide varying types of support to a patient with glioblastoma, then a small network, such as participant 11 may be insufficient to meet the full range of the patient’s needs (Figure 2). Conversely, members of a larger network may be able to share the responsibilities of supporting a patient, reducing the individual load of time, emotions, energy, and resources. A larger PSN could also distribute the emotional burden associated with supporting a glioblastoma patient, potentially reducing the likelihood of compassion fatigue among network members. This presents potential research opportunities, as a neurologic disability may restrict patients’ PSN over time. One area for further study could be to assess the PSNs of the primary caregiver, to determine whether their PSNs provide additional support and help mitigate compassion fatigue. Additionally, studying the PSN characteristics may inform differences in how social influences in cancer may be affected by ethnic groups or social institutions like church communities.²⁷

Future research should investigate whether certain PSN characteristics are associated with QOL so as to inform the PSN needs of patients. Analyzing personal characteristics of PSN alters, such as a larger network of daughters, as hinted in our qualitative study, could also shed light on the specific strengths or needs of a patient’s PSN. While there were some statistically significant correlations between EORTC-QLQ-BN20 categories and PERSNET characteristics, specifically hair loss and motor dysfunction, the small sample size of the study makes it difficult to identify the underlying mechanisms of these correlations. However, these findings do highlight the physical symptoms relevant to a patient’s social network and can inform which symptoms are worth tracking over time.

There were limitations to the current pilot project. Being a pilot study, we did not power the study to detect statistically significant differences in comparing network characteristics with QOL measures, tumor characteristics, demographics, and overall survival. However, the purpose was to determine trends in data and to inform future studies on the important variables to collect and use as a basis for power calculations. The current study provides the basis for the possibility of a larger trial studying how PSN may affect cancer treatment outcomes in this patient population, how PSN may change over the course of an illness with neurologic deterioration, and, therefore, where behavioral and social interventions may be most necessary during the evolution of illness.^9,18,20,22

Another limitation of the study is that surveys were conducted with both patients and their caregivers, but we do not know if the patient and caregiver would evaluate the patient’s PSN similarly if they were asked separately. Additional research is needed to measure coherence between patient and caregiver reports of a patient’s PSN, especially since communication may be challenging in this patient population where language and/or cognition can be affected early in the disease. There were some signals of possible differences in PSN by sex. While there were differences between social network characteristics in males and females, they were not statistically significant. This may be due to the fact that this was a pilot study and therefore, was not designed for statistical significance. These differences provide hypotheses to be tested in the next phase of the study.

This pilot study successfully demonstrated the feasibility of quantitatively analyzing the PSNs of glioblastoma patients. The findings underscore the need to develop validated measures for assessing social support structures, which could enhance supportive care trials and interventions by providing an additional measure of social support for glioblastoma patients. Future research should explore the impact of PSNs on QOL and outcomes such as overall survival, as well as how PSNs evolve over time in this patient population. Additionally, validating the correlation between patient and caregiver-reported PSNs will be crucial for those unable to communicate due to neurological impairments. This pilot project lays the groundwork for future research into the role of behavioral and social interventions in improving health outcomes.

Supplementary Material

Supplementary material is available online at Neuro-Oncology Practice (https://academic.oup.com/nop/).

Funding

This project was funded in kind through the Department of Neurology at Brigham and Women’s Hospital.

Conflict of interest statement

The authors declare that there are no conflicts of interest relevant to this work. No financial or non-financial relationships, activities, or affiliations with any organizations or entities influenced the design, execution, or interpretation of this research.

Authorship Statement

JYR, TB, and AD conceptualized the study, designed the methodology, and supervised the project. GY, TB, and AD helped conceptualized the study. CM, GY, and VN helped recruit patients. CM, PJM, and ZT conducted data collection and analysis. VAD, PJM, ZRR, ZT, TB, and AD provided critical feedback on the manuscript and assisted with data interpretation. JYR, PJM, and ZT drafted the manuscript and contributed to its revision. All authors reviewed the manuscript critically for important intellectual content and approved the final version for submission.

Data Availability

The data that support the findings of this study are available from the corresponding author, Dr. John Y. Rhee, john_rhee@dfci.harvard.edu, upon reasonable request. Due to privacy, some data may not be publicly available.