Bone-specific response according to MDA criteria predicts immunotherapy efficacy among advanced non-small cell lung cancer (NSCLC) patients

Purpose The presence of bone metastasis at baseline has been associated with dismal prognosis under immunotherapy in advanced non-small cell lung cancer (NSCLC). Response Evaluation Criteria in Solid Tumors (RECIST) criteria may be limited for bone-specific response evaluation. Whether their assessment through MD Anderson (MDA) criteria predict immunotherapy efficacy is unknown. Materials and methods We conducted a single-center retrospective study to assess the use of MDA criteria in evaluating bone metastasis in NSCLC treated with immunotherapy. Radiological imaging were reviewed to classify bone lesions as osteolytic, osteoblastic, or mixed. Bone response to treatment data was classified according to MDA criteria. Results 222 patients received single-agent immunotherapy. The presence of bone metastasis increased the risk of death both in the univariate (HR: 1.46, 95% CI, 1.05–2.03, p = 0.024) and in the multivariate model (HR: 1.61, 95% CI, 1.10–2.36, p = 0.015). According to MDA criteria, 57.3% of patients had progressive disease as best response, 29.5% stable disease, 11.4% partial response and 1.6% complete response. Bone-specific objective response was associated with a significantly increased median overall survival (11.3 vs. 3.1 months, p = 0.027) and longer median progression-free survival (6 vs. 2.1 months, p = 0.056). The median time to bone failure (TBF) was 2.4 months (IQR, 1.67–3.0). In 25.7% of cases, TBF was shorter than progression-free survival according to RECIST 1.1 criteria. TBF was positively correlated with overall survival (HR = 0.73, p = 0.00019). Conclusions MDA criteria represent a reliable tool in assessing bone-specific response, offering a more accurate evaluation with the aim to earlier predict survival outcomes or treatment failure compared to RECIST criteria for advanced NSCLC patients receiving immunotherapy. Supplementary Information The online version contains supplementary material available at 10.1007/s00432-022-04120-z.


Introduction
Advanced non-small cell lung cancer (NSCLC) represents the first cause of cancer-related death worldwide (Daniele et al. 2015). In the last decade, the switch from standardized platinum-based chemotherapy toward a biomarkerdriven treatment strategy has dramatically extended the life expectancy of NSCLC patients. Programmed cell death-1 (PD-1) and programmed cell death-ligand 1 (PD-L1) inhibitors demonstrated their superiority over chemotherapy in the non-oncogene addicted disease, either as single-agent therapy or combined with chemotherapy, depending on PD-L1 expression (Reck et al. 2016;Mok et al. 2019;Gandhi et al. 2018;Paz-Ares et al. 2018;Di Federico et al. 2021a, b). Andrea De Giglio and Chiara Deiana contributed equally to this work.
Besides PD-L1 expression, other clinical and biomolecular factors, such as the Eastern Cooperative Oncology Group Performance Status (ECOG PS) at diagnosis, the presence of concurrent mutations in specific genes (e.g., STK11 and KEAP1), and location of metastases have been proposed as predictors of response to immunotherapy (Facchinetti et al. 2020;Di Federico et al. 2021a, b;Lindblad et al. 2021). In addition, the tumor microenvironment (TME) may determine different overall or site-specific responses to immunotherapy (Oliver et al. 2018).
Bone represents one of the most frequent metastatic sites of lung malignancies, with an estimated incidence of 30-40% of all patients with NSCLC (Riihimäki et al. 2014). Of all patients with bone lesions, in 60% of cases these metastases are already present at first diagnosis, while in the other 40% they appear in the next 9 months (Daniele et al. 2015). Bone involvement has been associated with poor survival either with platinum-based chemotherapy or immunotherapy in NSCLC patients (Qin et al. 2021;Tournoy et al. 2018;Landi et al. 2019).
Tumor response is widely assessed according to the Response Evaluation Criteria in Solid Tumors (RECIST) (Eisenhauer et al. 2009). However, both RECIST version 1.1 and iRECIST consider bone metastasis as target lesion only in lytic or mixed lytic blastic lesion and with a soft tissue component of at least 10 mm. Therefore, purely osteoblastic or bone lesions with a small soft tissue component cannot be measured with such criteria. Nonetheless, a quality evaluation can be performed for non-target lesions: complete response in case of the disappearance of all lesions and normalization of the tumor marker level, non-complete response or non-progressive disease in case of persistence of one or more non-target lesions and/or presence of tumor marker level above the standard threshold, and progressive disease in case of unequivocal progression or the appearance of new lesions (Eisenhauer et al. 2009;Seymour et al. 2017).
The MD Anderson (MDA) criteria offer a more comprehensive evaluation of bone lesions. The MD Anderson (MDA) criteria offer a more comprehensive evaluation of bone lesions. In fact all bone lesions, including those classified as not target lesions by the RECIST criteria, such as purely osteoblastic lesions and lytic lesions without a soft tissue component, are included in the assessment. Furthermore, what is regarded as response in the MDA criteria includes the disappearance of the lesion for osteoblastic metastases and a qualitative change for lytic lesions, such as the appearance of sclerosis (Hamaoka et al. 2004). Thus, the MDA criteria evaluate bone-specific response to treatment for all patients with bone metastases, as they are designed to include all types of lesions and assess the various changes associated with treatment.
The current work aimed to explore the use of MDA criteria in the evaluation of response in NSCLC patients with bone metastasis. In addition, we investigated whether qualitative differences in bone lesions could affect the response to immunotherapy.

Materials and methods
We conducted a retrospective, observational study including all consecutive patients affected by advanced NSCLC and treated with single-agent immunotherapy between 2015 and 2021 at the Sant'Orsola-Malpighi University Hospital (Bologna, Italy). We extracted clinical and biological data from medical records. The following variables have been collected: age, gender, tumor histology, smoking status, PD-L1 expression, Eastern Cooperative Oncology Group (ECOG) performance status (PS) at baseline, anticancer treatments, radiological findings at baseline and during the follow-up, last follow-up, cause and date of death.
Two physicians (CD, ADG) independently reviewed radiological imaging of patients presenting bone metastasis at diagnosis, including CT scans and PET with low dose CT scans. Bone lesions were classified as osteolytic, osteoblastic, or mixed-type if both components were present. Bone response to treatments data was collected and classified according to the MDA criteria: osteoblastic lesions were classified as responding to treatment if they decreased in size (PR) or completely disappeared (CR), while lytic lesions were deemed in response if a sclerotic rim appeared (PR) or if they had a complete sclerotic fill-in (CR) (Hamaoka et al. 2004).
In Fig. 1S we provided an example of response evaluation according to the MDA criteria.
After appropriate approval from an Internal Independent Ethics Committee (approval no. 2381/2019), we conducted this study following the Declaration of Helsinki (1964).

Statistical methods
Continuous and categorical variables were described as median values and proportions. T-test (or ANOVA, or Pearson correlation test if needed) and Chi-Squared test (or Fisher's exact test, if needed) were performed to compare means and proportions. Shapiro test was performed to verify the normality of data distribution for each variable of interest.
Overall survival (OS) was defined as the time from treatment start to death from any cause and represented the primary endpoint. Progression-free survival (PFS) was defined as the time occurring from treatment start to the first radiological or clinical disease progression, or death from any cause. Time to bone failure (TBF) was defined as the time occurring from treatment start to first radiological or clinical bone disease progression or death from any cause.
The overall response was defined as a partial or complete response to treatment according to RECIST 1.1 criteria. The bone objective response was described as a partial or complete response of bone lesions to treatment according to RECIST 1.1 or MDA criteria.
Patients still alive at data cut-off (July 2021) were censored at last contact. The Kaplan-Meier method was used to estimate median survival times. The Log Rank Test was used to compare survival outcomes. The reverse Kaplan-Meier method was adopted to calculate the median time of followup. A Cox regression model was performed to explore the relationship between clinical or biological variables and survival outcomes. First, a univariate analysis was performed for both survival endpoints; then, variables reaching a p-value < 0.1 or considered clinically relevant were included in a multivariable model. A p-value ≤ 0.05 was considered statistically significant. Statistical analyses were performed with R-Studio version 1.4.1717, using the following packages: "dplyr", "prodlim", "survminer", "survMisc".

Demographic analysis
A total of 222 patients received single-agent immunotherapy at our institution between March 2015 and June 2021. The median age was 69.5 years (IQR, 63.7-75.1). 61.7% of patients were male, 76.1% had non-squamous histology, 68.1% had a smoking history, and 84% had an ECOG PS of 0 or 1. 50.7% of patients had more than two metastatic sites before the start of immunotherapy. 18.9% and 12.6% of patients showed liver or brain involvement, respectively. 27.5% of patients had ≥ 1 bone metastasis at immunotherapy baseline. Of them, 14.4% were osteolytic, 5.9% were osteoblastic, and 7.2% were mixed-type. Baseline characteristics showed no relevant distribution imbalances, except for a significantly higher prevalence of ≥ 2 metastatic sites among patients with bone metastasis (Table 1).
The presence of bone metastasis was associated with an increased risk of death either in the univariate model (HR: 1.46, 95% CI, 1.05-2.03, p = 0.024) or in the multivariate models adjusting for age, histology, number of metastatic sites, line of treatment, PD-L1 expression, brain and liver sites of metastasis (HR: 1.61, 95% CI, 1.10-2.36, p = 0.015). Within the same model, the presence of liver metastases at baseline was significantly associated with reduced survival (HR: 1.66, 95% CI, 1.12-2.46, p = 0.012) ( Table 2).
Analyzing bone responders' biological and clinical characteristics, we did not find any correlation within the univariate model ( Table 5).
The median time to bone response was 2.7 months (IQR, 2.5-4.1). On the other hand, the median time to bone failure (TBF) was 2.4 months (IQR, 1.67-3.0). Considering patients with bone PD as best response to immunotherapy, we found that in 9/35 (25.7%) cases the TBF was shorter than PFS according to RECIST 1.1 criteria. The TBF was positively correlated with OS (HR = 0.73, p = 0.00019), as shown in (Fig. 5).
Analyzing the same bone lesions according to RECIST criteria, we found out that 10 (16.4%) were target lesions, 27 (44.3%) were considered non-target lesions, and 24 (39.3) were not evaluable due to the chosen evaluation method (CT without contrast, PET with low dose CT scan).
In addition, 12 patients (19.6%) received a palliative course of radiotherapy for pain control (single fraction, 8 Gy). No prophylactic surgery was performed.

Discussion
We conducted a single-center retrospective study on 222 patients affected by advanced NSCLC, investigating the role of bone-specific response as a predictor of the efficacy of single-agent immunotherapy. Previous studies that analyzed the prognostic role of bone metastasis in NSCLC patients consistently showed decreased OS in patients with bone lesions at baseline, as compared with those without bone lesions. In a study by Qin et al. (2021) (Zhang et al. 2017). Our data fit in line with these studies, showing an increased risk of death in patients with bone metastasis at diagnosis and reinforcing the internal validity of the subsequent analyses. We then analyzed whether distinct types of bone metastases at baseline correlate with a different response to treatment. Bone metastases in patients with lung cancer are usually lytic, although mixed or osteoblastic morphologies are also observed. Distinct patterns of cytokines underlie the development of different types of lesions, according to the balance between bone formation and resorption (Wang et al. 2020). Our data showed that the type of bone metastasis does not influence the OS. Furthermore, we assessed bone-specific responses using the MDA criteria. Our data showed that the bone-specific response assessed by MDA criteria was significantly correlated with survival outcomes. Consistently, a retrospective experience including 16 NSCLC patients treated with nivolumab evidenced that early response evaluated with MDA criteria may be a predictor of prognosis and of disease response evaluated with RECIST 1.1 criteria (Nakata et al. 2020).
The small number of patients constituted a relevant criticism of their work, impeding an affordable multivariate assessment. In addition, the authors recognized the short median follow-up time (12.2 months) as a limitation of their work. Conversely, our analysis's median follow-up time was longer (30.1 months), probably due to the inclusion of patients treated with upfront immunotherapy.
We demonstrated that patients experiencing a bone-specific objective response (PR/CR) had longer median OS and PFS. On the other hand, a shorter TBF predicted an overall systemic treatment failure and increased risk of death, as bone PD according to MDA criteria preceded systemic disease progression in approximately 1/4 of cases. The correlation between bone-specific response and outcome has already been explored in oncogene-addicted NSCLC, and osteoblastic reactions in patients treated with EGFR inhibitors have been associated with favorable outcomes (Pluquet et al. 2010).
However, this is to our knowledge the first study that demonstrated a statistically significant correlation between bonespecific response and survival in non-oncogene addicted NSCLC patients treated with immune-checkpoint inhibitors.
It is worth noting that almost 40% of patients had bone lesions that were not evaluable with the RECIST 1.1 criteria. Notably, the response evaluation with methods different from CT scan with contrast medium, such as CT without contrast or PET with low dose CT scan, can be frequent in   -PD  SD  PD  3  Non-target  Non-CR/Non-PD  PD  SD  4  Target  PD  PD  PD  5 Non-target PD PD PD 6 Not evaluable Not evaluable SD PD 7 Non-target Non-CR/Non-PD PR PD 8 Non-target PD PD PD 9 Non-target PD PD PD 10 Non-target Non-CR/Non-PD SD PD 11 Non-target PD PD PD 12 Not evaluable PD PD PD 13 Target  PR  PR  PR  14  Not evaluable  PD  PD  PD  15  Not evaluable  PD  PD  PD  16  Not evaluable  Not evaluable  PD  PD  17  Not evaluable  Not evaluable  PD  PD  18  Non-target  PD  PD  PD  19  Not evaluable  Not evaluable  PD  PD  20  Non-target  PD  PD  PD  21  Target  PD  PD  PD  22 Non-target PD PD SD 23 Non-target Non-CR/Non-PD SD PD 24 Non-target Non-CR/Non-PD SD PD 25 Not evaluable  Not evaluable  PD  PD  26  Non-target  PD  PD  PD  27  Not evaluable  PD  PD  PD  28  Not evaluable  Not evaluable  PD  PD  29 Non-target Non-CR/ Non-PD  SD  PD  30  Target  PR  PR  PR  31 Non-target Non-CR/Non-PD PR SD 32 Target  SD  SD  PD  33 Non-target Non-CR/Non-PD SD SD 34 Target  PD  PD  PD  35 Non-target  Non-target Non-CR/Non-PD SD PD

Fig. 3
Overall survival according to bone-specific response assessed with MDA criteria clinical practice. The MDA criteria can be a useful integrating tool to categorize response to therapy in these settings. Finally, we did not find any correlation between the use of zoledronic acid and survival outcomes or bone-specific response according to MDA criteria, consistently with most studies evaluating their impact on survival in cancer patients (Henry et al. 2011;Scagliotti et al. 2012). However it should be noted that our study did not evaluate the occurrence of adverse skeletal events in relation with the use of zoledronic acid and thus the impact of this drug on prognosis has not been fully explored.
The main limitation of this work is represented by its retrospective nature and the limited sample size of patients included typically linked to a monocentric experience.
In addition, the inclusion of patients who underwent multiple lines of treatments may have affected the reliability of the analysis about the bone response, even if we preliminary considered the line of treatment within the multivariate analysis confirming the negative prognostic role of bone metastasis.
Moreover, our analysis did not explore the prognostic value of number, size or impending fracture of bone metastasis.
The investigator-related evaluation of disease progression constituted another criticism of our investigation. Nevertheless, two physicians have independently reviewed the radiological findings a posteriori. Finally, the correlation between the time to bone failure and overall survival may have been biased by the immortal time bias, albeit the median time to response and median time to bone failure were shorter than 3 months, thus considerably reducing this risk.
Overall, this might be a valid starting point for further studies analyzing the prognostic nature of bone-specific response assessed with the MDA criteria.

Conclusion
MDA criteria represent a feasible and reliable tool in assessing bone-specific response to immunotherapy in advanced NSCLC, offering a more accurate evaluation and additional information capable of an earlier prediction of longer survival or treatment failure compared to RECIST 1.1 or iRECIST. Thus, we propose the inclusion of MDA criteria in the response assessment of future clinical trial  testing immunotherapy strategies in patients with advanced NSCLC. Further studies will evaluate the consistency of our findings in NSCLC patients treated with first-line chemoimmunotherapy.
Author contributions All authors contributed to the study conception and design. Material preparation and data collection were performed by ADeG, CD and ADiF. Analysis were performed by ADeG. The first draft of the manuscript was written by CD and ADeG. Review and editing was performed by ADiF. All authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.
Funding Open access funding provided by Alma Mater Studiorum -Università di Bologna within the CRUI-CARE Agreement. The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Conflict of interest
The authors have no relevant financial or non-financial interests to disclose.

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethics approval
This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Internal Ethics Committee (approval no. 2381/2019).

Consent to participate
Informed consent was obtained from all individual participants included in this observational study when possible.
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 http://creativecommons.org/licenses/by/4.0/.