Antibiotics and steroids, the double enemies of anticancer immunotherapy: a review of the literature

Abstract

The advent of immunotherapy in onco-haematology has represented a kind of revolution that has been able to modify the prognosis of numerous tumours that until recently would have been rapidly lethal. While much is known about the mechanism of action of these drugs, relatively little is known about the factors that represent potential predictors of response and toxicity. Among these factors, the simultaneous administration of antibiotics and/or steroids seems to have a negative impact. Furthermore, several retrospective studies have highlighted the strong link between cancer and gut microbiota, regardless of the tumour site, and how microbiota, playing a key role in the prevention of systemic inflammation at various levels and in the intestinal homeostasis, can be negatively influenced by the dysbiosis caused by antibiotic therapy administered during or in the weeks immediately preceding the start of immunotherapy. Moreover, we assume that the concurrent administration of steroids, which is often necessary in cancer patients, likely results in a deleterious effect on the therapeutic outcomes of immunotherapy. In this review, we will try to clarify the evidence on the possible detrimental effects of antibiotics and steroids, which are currently considered the double enemies of anticancer immunotherapy.

Introduction

The advent and development of new cancer treatments such as immunotherapy (IO) have led to significant improvements, especially for some patients who have been able to benefit from or experience dramatic and lasting responses, even in advanced cancers that until a few years ago were considered to have a bad prognosis with extremely short survival times. The main mechanism of action of the immune checkpoint inhibitors (ICIs) consists of reversing the blockade of the inhibitory effect between the T lymphocyte receptor and its ligand located on neoplastic cells or myeloid cells, thus allowing the T lymphocyte-mediated response against tumour antigens. ICIs are monoclonal antibodies (mAbs) that target PD-1 (programmed cell death-1) and PD-L1 (programmed cell death ligand-1) or target cytotoxic T-lymphocyte antigen 4 (CTLA-4), which act to enable cytotoxic lymphocytes to regain their ability to attack tumour cells [1]. However, the magnitude of the responses and the spectrum of toxicities to IO are still extremely variable, and further research is needed to identify more reliable predictive factors, particularly those concerning the host. A refractoriness to ICIs is present, depending on study, in over 1/3 of treated patients, and often the main causes remain obscure [2]. Among the predictors of response to treatment with ICIs, the levels of PD-L1, tumour mutation burden (TMB), gene expression signatures of inflammation, neutrophil-to-lymphocyte ratio (NLR), and state of microsatellites at the baseline were recently identified [3–8]. However, in this context, the most promising would seem to be represented today by the modulation of gut microbiota by antibiotic therapy (ATB). Understanding the impact of the gut microbiome on immune function may be the key to developing consistently effective immunotherapy. It is now an accepted concept that, in the context of induced dysbiosis, the gut microbiota is able to alter the systemic response of the immune system, potentially contributing to the onset of chronic inflammatory disorders such as obesity, Crohn's disease and diabetes mellitus type 2 [9]. Therefore, gut microbiota is able to play a key role, even in the context of the responses to IO. Not surprisingly, this approach revealed that the gut microbiota from patients (pts) who failed to benefit from ICIs differed in its composition from those who responded [10–20]. A similar potential detrimental effect is observed, but through a very different mechanism of action, as result of a concurrent treatment with steroids and IO. Steroids are often necessarily administered to cancer pts to relieve symptoms and have been able to reduce the response to IO, as demonstrated by overall response rate (ORR), progression free survival (PFS) and overall survival (OS). In this review, we try to clarify the current evidence on the possible detrimental effects of the associations between ATB and IO and between steroids and IO.

Antibiotics and immunotherapy

Putative mechanism of the detrimental effect

Tumours, especially at an advanced stage, are often accompanied by inflammation and infections that frequently require ATB, according to the guidelines [15]. It is known that the gut microbiota is implicated in numerous physiological and pathological processes in humans and is closely related to a good functioning of the immune system. Several preclinical studies have highlighted the role of the gut microbiota in regulating responses to IO [17, 18]. First, in mouse models, the lack of an intact gut microbiota (in germ-free or ATB-treated mice) dramatically decreased the antitumour activity of ICIs induced by CD4-positive T cells. As a result, the faecal microbiota transplantation in these germ-free mice from pts treated with immune checkpoint blockade restored the activity of ICIs [27]. Moreover, observational studies showed the relevance of geographical locations, differences in the composition of the gut microbiota and responses to the IO [19, 20, 27]. ATBs can induce intestinal dysbiosis through the diminution of the abundance of critical species or through the complete eradication of important commensals, therefore, altering normal gut homeostasis with effects on local and whole-body metabolism, as well as having major consequences for the immune system. Several retrospective studies have highlighted the strong link between cancer development and gut microbiota [10–21, 23], regardless of the primary tumour site, and how microbiota, playing a key role in the prevention of systemic inflammation at various levels and in the intestinal homeostasis, can be negatively influenced by the dysbiosis caused by ATB administered during or in the weeks immediately preceding the start of IO [10, 24–30, 30, 33–35]. Recently, some studies showed the possible promoting effect of a previous ATB on the onset of cancer. This result could be explained merely by modulation of the intestinal microbiota. Bacteria induce carcinogenesis through two main mechanisms: the induction of chronic inflammatory processes leading to cancer in various human organs, or the production of carcinogenic metabolites. A recent large meta-analysis by Petrelli et al. has shown that the use of ATB represents an independent risk factor for the development of cancer (OR 1.18, 95% CI 1.12–1.24, p < 0.001), especially for lung, kidney and pancreatic cancer, as well as for some lymphomas and multiple myeloma [21]. A similar meta-analysis by Zhang et al. showed that the administration of oral broad-spectrum ATB may be associated with an increased risk of colon cancer, especially in the right colon, probably due to the differences in gut microbiota and carcinogenesis mechanisms [23].

Clinical evidence (Table 1)

Table 1.

Retrospective clinical trials addressing the impact of ATB on the activity of ICIs

Author (ref)	Primary tumor site (number of evaluable pts)	Antibiotic window	Outcome (p-values)
Galli [29]	NSCLC (157)	1 month PRE or 3 mos POST	↓ PFS (0.0001) and ↓ OS (0.0004)
Huemer [30]	NSCLC (30)	1 month PRE or 1 month POST	↓ PFS (0.031) and ↓ OS (0.021)
Lurienne [34]	Multiple Sites (2208 for PFS and 5560 for OS)	2 mos PRE or 2 mos POST	↓ PFS and ↓ OS
Thompson [61]	NSCLC (74)	6 wks PRE	↓ PFS (0.02) and ↓ OS (0.004)
Rubio [62]	NSCLC (168)	2 mos PRE or 1 month POST	↓ PFS (0.028) and ↓ OS (0.026)
Ouaknine [63]	NSCLC (72)	2 mos PRE or 1 month POST	↓ OS (0.03)
Do [64]	NSCLC (109)	1 month PRE or 1 month POST	↓ OS (0.0004)
Zhao [65]	NSCLC (109)	1 month PRE or 1 month POST	↓ PFS (0.0001) and ↓ OS (0.0021)
Schett [66]	NSCLC (218)	2 mos PRE	↓ PFS (0.001) and ↓ OS (0.001)
Lalani [67]	RCC (146)	2 mos PRE or 1 month POST	↓ PFS (0.08) and ↓ ORR (0.026)
Mohiuddin [35]	Melanoma (114)	45 days PRE or 45 days POST	↓ OS (0.001)
Elkrief [68]	Melanoma (74)	30 days PRE	↓ PFS (0.01) and ↓ ORR (0.01)
Tinsley [24]	NSCLC (58), RCC (46), Melanoma (201)	14 days PRE or 42 days POST	↓ PFS (0.049) and ↓ OS (0.001)
Routy [27]	NSCLC (140), RCC (67), UC (32)	2 mos PRE or 1 month POST	↓ PFS (0.017) and ↓ OS (0.001)
Derosa [28]	NSCLC (249), RCC (121)	1 month PRE	↓ PFS (0.001) and ↓ OS (0.001)
Pinato [30]	NSCLC (119), Melanoma (38), Other (39)	30 days PRE or concurrently	↓ OS (0.001) and ↓ ORR (0.01) (PRE)
Masini [69]	NSCLC (78), Melanoma (57), RCC (29), Other (5)	Any time	↑ OS (0.028)
Kulkarni [70]	NSCLC* (148), RCC** (55)	1 month PRE or any time POST	↑ PFS (0.01)*, ↑ OS (0.01)*, ↑ ORR (0,01)* ↓ PFS (0.04)**

Ref reference, pts patients, NSCLC non-small-cell lung cancer, RCC renal cell carcinoma, UC urothelial cancer, ATB antibiotic therapy, ICIs immune checkpoint inhibitors, PFS progression free survival, ORR overall response rate, OS overall survival, mos months

The main evidence derived from researches reported over the last 10 years confirms that ATB exposure tends to increase cancer risk and, unfortunately, that it reduces the efficacy of cancer therapies. Exposure to a broad-spectrum ATB has been shown to be able to negatively influence the results of IO with ICIs by the modulation of gut microbiota. Several studies have highlighted this close correlation between ATB and ICIs [2, 10, 24–30, 30, 33–35]. However, to date, little is known about the potentially detrimental timing of ATB administration and the start of IO. A recently published retrospective study by Tinsley et al. demonstrated that cumulative or prolonged use of ATB is an independent negative predictor of PFS and OS in pts with advanced cancer treated with ICIs [24]. In a prospective study, Pinato et al. evaluated the correlation of ATB with ORR and OS in approximately 200 cancer pts treated with ICIs. This study showed that there is a statistically negative correlation for all outcomes in the case of ATB administration before the beginning of treatment with ICIs but not in the case of the concurrent administration of ATB and IO [30]. Derosa et al. examined 121 pts with advanced renal cell carcinoma (RCC) and 239 with non-small cell lung cancer (NSCLC) treated with anti-PD-L1 mAbs. Those receiving ATB within 30 days of beginning ICIs were compared with those who did not. ATB administration was associated with an increased risk of primary progressive disease and shorter PFS and OS. The same trend was observed if ATB was administered within 60 days of starting ICI [2]. These findings suggest a time-dependent and partial repopulation of the gut microbiota after ATB discontinuation. Some studies have shown that it requires from 1 to 3 months after the suspension of the ATB for the microbiota to return to the initial situation, but it sometimes requires years to obtain a full recovery [31, 32]. Therefore, in the current state of knowledge, what has been highlighted with certainty is the existence of a correlation between ATB and a poor outcome in pts treated with ICIs. However, it is debated whether ATB use represents a true independent therapeutic biomarker or is a surrogate for pts with a worse overall prognosis (pts who required ATB use often had additional comorbidities). Wilson et al. recently published a systematic review and large meta-analyses of 766 observational studies on ATB and ICIs. Eighteen studies and 826 pts met the inclusion criteria for analyses. OS was 3.4 times longer in pts who did not receive any ATB in the 42 days prior to IO (HR 3.43, 95% CI 2.29–5.14, p < 0.0001), and PFS was also longer in pts who did not receive ATB (pooled HR 1.65, 95% CI 1.3–2.1, p < 0.0001). Conversely, exposure to ATB before 60 days starting or during IO seems not to influence the clinical outcomes of IO [33]. Moreover, two further reviews and meta-analyses of Lurienne et al. [34] and Mohiuddin et al. [35] explored and confirmed this detrimental link between the time of ATB exposure and the start of IO both in NSCLC and melanoma pts, respectively. Finally, it is known that there are still several obscure points, such as the time frame at which ATB administration is more deleterious for IO, the basal state of the gut microbiota and the possibility of intervening with treatments such as faecal transplantation [36, 37], diets, prebiotics and probiotics aimed at preventing and remediating ATB-induced dysbiosis. Therefore, manipulation of the microbiota for therapeutic benefits is a major goal. Further prospective studies are warranted to confirm these findings.

Steroids and immunotherapy

Mechanism of the detrimental effect

As a consequence of the well-known immunosuppressive effect of corticosteroids, today these drugs are considered able to reduce the efficacy of immunotherapy, and the guidelines report the use of simultaneous steroids as a relative contraindication. Giles et al. showed that the immunosuppressive effect of dexamethasone (DEX) consisted of upregulation of CTLA-4 mRNA and protein in CD4 and CD8 T cells and blocked CD28-mediated cell cycle entry and differentiation. Therefore, it seems that CTLA-4, but not PD-1 blockade, can partially prevent some of the negative effects of DEX on the immune response [38, 39]. Corticosteroids perform their immunosuppressive activity by inhibition of certain priority T-lymphocyte functions by suppressing IL-2-mediated activation of effector T cells [40] and the increase in regulatory T lymphocytes [41]. Furthermore, it seems that steroids are able to cause the alteration of the microbiome [42] and stimulate M2 polarization of macrophages [43]. However, their use and the clinical implications of an interaction between IO and steroids remain unclear. Pts with refractory anorexia, nausea, fatigue, pain, brain metastases, and dyspnoea require care and are often given corticosteroids as the main treatment for symptom palliation. Moreover, ICIs cause adverse events (irAEs) that can involve any organ, and steroid use represents the first antidote.

Clinical evidence

The relationship between steroids and IO remains very controversial. Because of their immunosuppressive properties, corticosteroids are both the principal treatment of irAEs due to ICIs [44] and an exclusion criterion for ICIs clinical trials considering a threshold of ≥ 10 mg of prednisone equivalent daily the usual cutoff [45, 46]. Usually, when irAEs are grade 2 or higher, steroid administration such as prednisone is used at a dosage of 0.5–2 mg/kg/day with a possible subsequent tapering of the induction dose if the symptoms resolve [47]. This management of irAEs does not seem to have consequences on the efficacy of IO [48] or the retreatment with ICIs after serious irAEs [49–51] or significantly suppress the function of T cells [52, 53]. A large systematic review by Garant et al. showed that there was no objective data on the exact types of corticosteroids and the dose threshold above which an interaction could be measured clinically [54]. However, Arbor et al. demonstrated that pts on IO and who were taking 10 mg/day or more prednisone when they initiated ICIs for advanced NSCLC reported a significantly inferior objective RR, PFS and/or OS compared with those who used less than 10 mg/day of prednisone or the equivalent [55]. Doses > 10 mg of prednisone daily seem related to increasing infection rates in long-term steroids treated pts [56], and are considered immunosuppressive [57]. Another unknown point is the duration of simultaneous treatment with steroids. In a retrospective study recently published by Pan et al., it seems that prednisone > 10 mg/day for > 2 weeks may be associated with poorer survival outcomes [58]. De Giglio et al. found that early introduction of steroid therapy (≥ 10 mg of prednisone or equivalent) in 49 advanced NSCLC pts led to significantly poorer outcomes if steroids were administered for cancer-related symptoms but did not have a detrimental impact on prognosis if the steroids were started for other indications such as irAEs or COPD exacerbation [59]. In a large retrospective observational study of 2213 pts with NSCLC, advanced melanoma and advanced urothelial cancer, Drakaki et al. showed that pts taking baseline corticosteroids (19–30%) were more likely to have stage IV disease at diagnosis, brain and/or liver metastases and poorer ECOG PS scores at baseline. The administration of baseline corticosteroids was related to a 23–47% higher risk of death when compared with no use in the multivariable models (Table 2) [60].

Table 2.

Pts characteristics by baseline CS use, and OS HR by CS use (modified from Drakaki et al. [60])

	NSCLC (n = 862)		Melanoma (n = 742)		UC (n = 609)
	CS (n = 258)	No CS (n = 604)	CS (n = 182)	No CS (n = 560)	CS (n = 116)	No CS (n = 493)
Stage IV at diagnosis	74	61	34	29	42	35
ECOG PS ≥ 2 at ICI start	17	15	9	11	34	21
Site of metastases
Brain	26	19	31	21	5	2
Multivariable OS, HR (95% CI), CS use vs. no CS use (reference)
Model 1	1.35 (1.12, 1.62)		1.23 (0.97, 1.57)		1.47 (1.14, 1.90)
Model 2	1.34 (1.12, 1.61)		1.24 (0.97, 1.57)		1.44 (1.12, 1.87)

Multivariable models adjusted for age at ICI start, stage at diagnosis, race/ethnicity, ECOG PS and Charlson Comorbidity Index at ICI start; treatment sequence, brain metastases at ICI start, smoking status (NSCLC, UC), histology (NSCLC), grade (UC) in model 1 and prior steroid use in model 2

Pts patients, CS corticosteroids, NSCLC non-small cell lung cancer, UC urothelial cancer, ICI immune checkpoint inhibitors, ECOG PS Eastern Cooperative Oncology Group Performance Status, HR hazard ratio

Discussion

IO has led to drastic positive changes in the prognosis of a broad range of advanced tumours such as melanoma, NSCLC, and renal cell carcinoma, which had a poor prognosis until only a few years ago. Today, due to IO, we can truly discuss the “chronicity” of advanced cancer. However, only a small proportion of pts benefit significantly from this treatment. For this reason, future research will focus on identifying new predictive response factors and on defining the roles of the old ones. Knowledge resulting mainly from retrospective studies has taught us that the two well-identified "enemies" of IO are represented by ATB, mainly due to the dysbiosis they cause at the intestinal level, and by steroids administered concurrently with ICIs. Many points still remain obscure and are worthy of further research. The real mechanisms that can determine these negative interferences are not yet completely known, such as the timing of administration of ATB (e.g., how long before the start of IO or whether simultaneous administration is acceptable). Furthermore, does this concept apply only to some broad-spectrum ATB or could it be extended to all ATB? What are the components of the gut microbiota that are most negatively affected? Are there effective remedies to overcome this problem, such as faecal transplantation, diets, prebiotics, or probiotics? These findings reaffirm the importance for physicians to use ATB judiciously, avoiding the use of ATB unless for life-threatening conditions, especially in responding or stable pts during IO.

On the other hand, with IO widely becoming the standard of care for pts with various advanced cancers, an important pragmatic concern regarding the concurrent use of ICIs and corticosteroids has emerged. It seems clear to avoid steroid use if possible, but there are some good practical points to consider. First, in the case of symptomatic brain metastases, we should try to use the lowest effective dose of prednisone or equivalent as soon as possible. Second, in the case of irAEs, we should provide a quick tapering of induction dose if clinically indicated. Finally, we should try to avoid use of steroids for relief of symptoms such as pain, fatigue, appetite loss or paraneoplastic fever. The type and the allowed dosage of administered corticosteroids with concurrent IO (e.g., prednisone < 10 mg/day) remain open questions, as well as the question of whether a simultaneous administration of chemotherapy plus IO may overcome these negative influences.

Providing answers to these questions remains one of the main exciting challenges in the field of onco-haematology awaiting us in the coming years.